AWS Announces Availability of Amazon EC2 Hpc6a Instances for HPC

Amazon Web Services (AWS) has introduced Amazon Elastic Compute Cloud (Amazon EC2) Hpc6a instances, a new cloud instance type purpose-built for closely connected high performance computing (HPC) applications. These new cloud instances are now generally available to AWS users.

Hpc6a cloud instances, which are powered by AMD EPYC processors from the third generation, would enhance AWS’ HPC computing choices and provide “up to 65 percent” better pricing performance than equivalent compute-optimized Amazon EC2 instances already used for HPC applications.

AWS users can scale HPC clusters on AWS Cloud to run their most compute-intensive workloads like genomics, computational fluid dynamics, weather forecasting, molecular dynamics, computational chemistry, financial risk modeling, computer-aided engineering, and seismic imaging for even less money with Hpc6a instances. Hpc6a cloud instances are accessible on demand with no upfront obligations through a low-cost, pay-as-you-go use approach.

Low Latency, Low Jitter, Networking Capacity

Photo David Brown, Vice President of Amazon EC2 at AWS
“Purpose-built for HPC workloads, Hpc6a instances now help customers realize up to 65% better price performance for their HPC clusters at virtually any scale,” said David Brown, Vice President of Amazon EC2 at AWS.

As IT workloads further scale to solve increasingly difficult problems, clients are aiming to maximize pricing performance as they run HPC workloads that may expand to tens of thousands of servers on AWS, according to the hyperscaler.

The new Hpc6a cloud instances are designed to provide the best value for money when executing HPC applications at scale in the cloud, according to AWS. For HPC applications requiring complicated calculations, Hpc6a instances would provide “up to 65 percent” higher pricing performance across a variety of cluster sizes – up to tens of thousands of cores. Elastic Fabric Adapter (EFA), an Amazon EC2 network interface, is enabled by default on Hpc6a instances.

Hpc6a cloud instance clients may benefit from low latency, low jitter, and up to 100 Gbps of EFA networking capacity to boost operational efficiency and accelerate time-to-results for workloads that rely on inter-instance interactions using EFA networking. Hpc6a instances are powered by AMD EPYC 3rd Gen CPUs with up to 3.6 GHz clock speeds and 384 GB of RAM.

“We are excited to continue our momentum with AWS and provide their customers with this new, powerful instance for high performance computing workloads,” said Dan McNamara, Senior Vice President and General Manager, Server Business at AMD. “AMD EPYC processors are helping customers of all sizes solve some of their biggest and most complex problems.

From small universities to enterprises to large research facilities, Hpc6a instances powered by 3rd Gen AMD EPYC processors open up the world of powerful HPC performance with cloud scalability to more customers around the world.”

ParallelCluster, Nitro System

Clients can deploy Hpc6a cloud instances alongside other instance types using AWS ParallelCluster (an open-source cluster management tool), allowing them the option to run multiple task types optimized for different instances inside the same HPC cluster. The AWS Nitro System, a set of building elements that offload many traditional virtualization activities to specialized hardware and software to achieve high performance, high availability, and better security while minimizing virtualization overhead, is available for HPC6a instances. On-Demand Instances, Reserved Instances, and Savings Plans are all available for HPC6a instances.

“By consistently innovating and creating new purpose-built Amazon EC2 instances for virtually every type of workload, AWS customers have realized huge price performance benefits for some of today’s most business-critical applications,” said David Brown, Vice President of Amazon EC2 at AWS.

“While high performance computing has helped solve some of the most difficult problems in science, engineering, and business, effectively running HPC workloads can be cost-prohibitive for many organizations. Purpose-built for HPC workloads, Hpc6a instances now help customers realize up to 65% better price performance for their HPC clusters at virtually any scale, so they can focus on solving the biggest problems that matter to them most without the cost barriers that exist today.”

HPC6a instances are available in AWS GovCloud (US-West) and US East (Ohio), with availability in more AWS Regions coming soon.