HPC Cloud Updates WE 01 Dec 2024

AWS

AWS CPU Compute Futures are finally here. They gave them a more boring name but that’s essentially what we’re talking about. Future dated EC2 Capacity Reservation.

Announcing future-dated Amazon EC2 On-Demand Capacity Reservations | Amazon Web Services

AWS EC2 Capacity Reservations empowers you to secure compute capacity for critical future workloads up to 120 days ahead, ensuring seamless performance during peak demand events like product launches or seasonal sales.

Amazon Web ServicesChanny Yun (윤석찬)

Request future dated Amazon EC2 Capacity Reservations - AWS

Discover more about what’s new at AWS with Request future dated Amazon EC2 Capacity Reservations

Amazon Web Services, Inc.

12 times faster throughput to GPUs? I can think of some AIs that might need that to train themselves

Amazon FSx for Lustre increases throughput to GPU instances by up to 12x | Amazon Web Services

Amazon FSx for Lustre now features Elastic Fabric Adapter and NVIDIA GPUDirect Storage for up to 12x higher throughput to GPUs, unlocking new possibilities in deep learning, autonomous vehicles, and HPC workloads.

Amazon Web ServicesDanilo Poccia

Amazon FSx for Lustre now supports Elastic Fabric Adapter and NVIDIA GPUDirect Storage - AWS

Discover more about what’s new at AWS with Amazon FSx for Lustre now supports Elastic Fabric Adapter and NVIDIA GPUDirect Storage

Amazon Web Services, Inc.

Valkey GLIDE now has availability zone awareness. That’s handy.

Valkey GLIDE 1.2 adds new features from Valkey 8.0, including AZ awareness - AWS

Discover more about what’s new at AWS with Valkey GLIDE 1.2 adds new features from Valkey 8.0, including AZ awareness

Amazon Web Services, Inc.

Bahrain gets EC2 r7g instances, c7g in Paris and Osaka and r8g in Mumbai

Amazon EC2 R7g instances are now available in AWS Middle East (Bahrain) region - AWS

Discover more about what’s new at AWS with Amazon EC2 R7g instances are now available in AWS Middle East (Bahrain) region

Amazon Web Services, Inc.

Amazon EC2 C7g instances are now available in additional regions - AWS

Discover more about what’s new at AWS with Amazon EC2 C7g instances are now available in additional regions

Amazon Web Services, Inc.

Amazon EC2 R8g instances now available in AWS Asia Pacific (Mumbai) - AWS

Discover more about what’s new at AWS with Amazon EC2 R8g instances now available in AWS Asia Pacific (Mumbai)

Amazon Web Services, Inc.

Azure

After a bumper set of releases last week I think the Azure teams are taking a well deserved rest. Not much of HPC interest this week.

If you’re running HPC workloads for EDA then this benchmarking exercise may be of interest:

Accelerating EDA workloads on Azure - Best Practice and benchmark on Intel EMR CPU

The article evaluates the performance of the latest Azure VMs using the 5th Gen Intel® Xeon® Platinum 8537C (Emerald Rapids) processor by comparing them to…

TECHCOMMUNITY.MICROSOFT.COMRaymondTsai

Google Cloud

Maybe not HPC in the traditional sense but if you’re running HPC on K8 this will certainly be of interest. Google Cloud now have support for upto 65,000 nodes in a cluster

Google Kubernetes Engine supports 65,000-node clusters | Google Cloud Blog

With support for 65,000-node clusters, Google Kubernetes Engine offers more than 10X larger scale than the other two largest public cloud providers.

Google Cloud