AWS integró DRA con Elastic Fabric Adapter en EKS. Desde K8s 1.34, podés asignar interfaces EFA con conciencia de topología para workloads de AI y HPC.

AWS announces Amazon EC2 Hpc8a instances, the next generation of high performance computing optimized instance, powered by 5th Gen AMD EPYC processors (formerly code named Turin). With a maximum frequency of 4.5GHz, Hpc8a instances deliver up to 40% higher performance and up to 25% better price performance compared to Hpc7a instances, helping customers accelerate compute-intensive workloads while optimizing costs. Built on the latest sixth-generation https://aws.amazon.com/ec2/nitro/, Hpc8a instances are designed for compute-intensive, latency-sensitive HPC workloads. They are ideal for tightly coupled applications such as computational fluid dynamics (CFD), weather forecasting, explicit finite element analysis (FEA), and multiphysics simulations that require fast inter-node communication and consistent high performance. Hpc8a instances feature 192 cores, 768 GiB memory and 300 Gbps of Elastic Fabric Adapter (EFA) network bandwidth, enabling fast, low-latency cluster scaling for large-scale HPC workloads. Compared to Hpc7a instances, Hpc8a instances also provide up to 42% higher memory bandwidth, further improving performance for memory-intensive simulations and scientific computing workloads. Hpc8a instances are available today in US East (Ohio) and Europe (Stockholm). Customers can purchase Hpc8a instances via Savings Plans or On-Demand instances. To get started, sign in to the https://console.aws.amazon.com/ec2/?trk=d8ec3b19-0f37-4f8c-8c12-189f913e205c&sc_channel=el. For more information visit the https://aws.amazon.com/ec2/instance-types/hpc8a/ or https://aws.amazon.com/blogs/aws/amazon-ec2-hpc8a-instances-powered-by-5th-gen-amd-epyc-processors-are-now-available.

AWS ParallelCluster 3.13 is now generally available. Key features of this release include support for Ubuntu 24.04, an updated Slurm version 24.05.07 and support for Elastic Fabric Adapter (EFA)-enabled Amazon FSx for Lustre filesystems on official ParallelCluster AMIs. You can use EFA with your FSx Lustre filesystems to achieve higher throughput and complete jobs faster, reducing overall costs. To get started with enabling EFA with FSx Lustre filesystems on your clusters, follow the tutorial in the ParallelCluster User Guide - https://docs.aws.amazon.com/parallelcluster/latest/ug/tutorial-efa-enabled-fsx-lustre.html. ParallelCluster is a fully-supported and maintained open-source cluster management tool that enables R&D customers and their IT administrators to operate high-performance computing (HPC) clusters on AWS. ParallelCluster is designed to automatically and securely provision cloud resources into elastically-scaling HPC clusters capable of running scientific and engineering workloads at scale on AWS. ParallelCluster is available at no additional charge in the AWS Regions listed https://docs.aws.amazon.com/parallelcluster/latest/ug/supported-regions.html, and you pay only for the AWS resources needed to run your applications. To learn more about launching HPC clusters on AWS, visit the ParallelCluster https://docs.aws.amazon.com/parallelcluster/latest/ug/tutorials-v3.html. To start using ParallelCluster, see the installation instructions for ParallelCluster https://docs.aws.amazon.com/parallelcluster/latest/ug/install-pcui-v3.html and https://docs.aws.amazon.com/parallelcluster/latest/ug/install-v3-parallelcluster.html.

AWS announces Amazon EC2 Hpc8a instances, the next generation of high performance computing optimized instance, powered by 5th Gen AMD EPYC processors (formerly code named Turin). With a maximum frequency of 4.5GHz, Hpc8a instances deliver up to 40% higher performance and up to 25% better price performance compared to Hpc7a instances, helping customers accelerate compute-intensive workloads while optimizing costs. Built on the latest sixth-generation AWS Nitro Cards, Hpc8a instances are designed for compute-intensive, latency-sensitive HPC workloads. They are ideal for tightly coupled applications such as computational fluid dynamics (CFD), weather forecasting, explicit finite element analysis (FEA), and multiphysics simulations that require fast inter-node communication and consistent high performance. Hpc8a instances feature 192 cores, 768 GiB memory and 300 Gbps of Elastic Fabric Adapter (EFA) network bandwidth, enabling fast, low-latency cluster scaling for large-scale HPC workloads. Compared to Hpc7a instances, Hpc8a instances also provide up to 42% higher memory bandwidth, further improving performance for memory-intensive simulations and scientific computing workloads. Hpc8a instances are available today in US East (Ohio) and Europe (Stockholm). Customers can purchase Hpc8a instances via Savings Plans or On-Demand instances. To get started, sign in to the AWS Management Console. For more information visit the Amazon EC2 Hpc8a instance page or AWS news blog.

AWS ParallelCluster 3.13 is now generally available. Key features of this release include support for Ubuntu 24.04, an updated Slurm version 24.05.07 and support for Elastic Fabric Adapter (EFA)-enabled Amazon FSx for Lustre filesystems on official ParallelCluster AMIs. You can use EFA with your FSx Lustre filesystems to achieve higher throughput and complete jobs faster, reducing overall costs. To get started with enabling EFA with FSx Lustre filesystems on your clusters, follow the tutorial in the ParallelCluster User Guide - Creating a cluster with an EFA-enabled FSx Lustre . ParallelCluster is a fully-supported and maintained open-source cluster management tool that enables R&D customers and their IT administrators to operate high-performance computing (HPC) clusters on AWS. ParallelCluster is designed to automatically and securely provision cloud resources into elastically-scaling HPC clusters capable of running scientific and engineering workloads at scale on AWS. ParallelCluster is available at no additional charge in the AWS Regions listed here, and you pay only for the AWS resources needed to run your applications. To learn more about launching HPC clusters on AWS, visit the ParallelCluster User Guide. To start using ParallelCluster, see the installation instructions for ParallelCluster UI and CLI.