Amazon EC2 Hpc6a Instances
HPC instances powered by 3rd generation AMD EPYC processors
Amazon EC2 Hpc6a instances offer the best price performance for compute-intensive high performance computing (HPC) workloads in Amazon EC2. Hpc6a instances offer up to 65% better price performance over comparable Amazon EC2 x-86 based compute-optimized instances. Elastic Fabric Adapter (EFA) , powered by the AWS Nitro System, is enabled to provide low latency 100 Gbps network bandwidth for inter-node communication. Hpc6a instances are powered by 3rd generation AMD EPYC processors. AWS elasticity and scalability combined with Hpc6a instances price performance let you scale workloads such as computational fluid dynamics, weather forecasting, and molecular dynamics, at the best price performance in Amazon EC2.

Benefits
Cost-effective
Hpc6a instances are purpose-built to cost effectively run HPC workloads on AWS. To help you reduce the cost of running your HPC workloads, Hpc6a instances are priced up to 70% less than similar compute-optimized, x86 based instances.
Fast results
With 100 Gbps Elastic Fabric Adapter (EFA) networking, you have access to a high-throughput, low latency, low jitter channel for tightly coupled workloads. EFA networking allows you to scale faster and reduce time-to-insight for your HPC workloads.
Operational efficiency
With AWS ParallelCluster you can provision Hpc6a instances alongside other instance types within the same cluster. This gives you the flexibility to run different workloads optimized for different instances within your cluster.
Maximized resource efficiency
Hpc6a instances are built on the AWS Nitro System, a combination of dedicated hardware and lightweight hypervisor which delivers practically all of the compute and memory resources of the host hardware to your instances for better overall performance and security.
Features
Powerful AMD performance
Hpc6a instances are powered by two 48-core 3rd generation AMD EPYC 7003 series processors built on 7nm process nodes for increased efficiency, and 384 GiB of memory, with a total of 96 cores (4 GiB of memory per core). This allows you to run the most demanding HPC workloads while taking advantage of the price performance offered by AMD.
High performance networking and storage
Latest generation AWS Nitro cards, and EFA deliver 100 Gbps of network throughput to interconnected nodes. You can also use Amazon FSx for Lustre for sub-millisecond latencies, and up to hundreds of gigabytes per second of throughput for storage.
Built on AWS Nitro System
The AWS Nitro System is a rich collection of building blocks that offloads many of the traditional virtualization functions to dedicated hardware and software to deliver high performance, high availability, and high security while also reducing virtualization overhead.
Customers and Partners
"Using AWS, we quickly scaled and added the needed on-demand compute power for the CFD team, compared with the months required and significant capital to build and scale an on-premises HPC cluster.”
Colin Haubrich, Head of IT, Wisk Aero

"AWS has been a long-time partner supporting our business in many ways. The recent HPC solution has provided a step-change improvement in both speed and cost for some of Blue Origin’s more expensive computational workloads. Not only did they advance the AWS infrastructure, they also provided a top-notch team to help Blue Origin implement the change to realize cost savings and speed up."
John Paris, Senior Propulsion Engineer, Blue Origin

"TotalCAE Platform makes it easy for computer-aided engineering (CAE) departments to adopt the agility and flexibility of AWS in just a few clicks for hundreds of engineering applications like Ansys Fluent, Siemens Simcenter STAR-CCM+, and Dassault Systemes Abaqus. As an AWS HPC competency partner, we help customers run their CAE workloads in the cloud. With Hpc6a instances, we have seen up to a 30% performance boost for Computation Fluid Dynamics (CFD) workloads at a lower cost, enabling TotalCAE to offer customers industry leading price performance, and scalability in the cloud.”
Rod Mach, President, TotalCAE

“The release of the Amazon EC2 Hpc6a instance is yet another exciting announcement from AWS that enables Maxar to continue to meet and exceed our customer requirements for Big Compute workflows – whether to accelerate the research and operations of Numerical Weather Prediction (NWP) workloads or to create the world’s best, most up to date, and accurate Digital Twin models with our Maxar Precision3D product suite. The Hpc6a, featuring AMD EPYC 3rd generation processors, combined with the EFA networking capability provides us a 60% performance improvement over alternatives, while also being more cost efficient. This enables Maxar to strategically choose among the suite of AWS HPC cluster configurations that we’ve developed to best suit our clients’ needs while maximizing flexibility and resiliency.”
Dan Nord, SVP and Chief Product Officer at Maxar Technologies

“The Hpc6a combines the leading-edge interconnect technology of Elastic Fabric Adapter with the extremely versatile AMD EPYC 3rd generation chipset, providing a performance sweet spot for applications ranging from computational fluid dynamics, which relies on scalable hardware, to structural finite element analysis, which benefits from the high memory performance of AMD 3rd generation EPYC processors. Rescale’s customers solve some of today’s most critical challenges in areas ranging from autonomous vehicles to climate to drug discovery using HPC, and with the Hpc6a we’re excited to give them another powerful tool for accelerating innovation and scientific progress."
Mulyanto Poort, VP of High Performance Computing at Rescale

“Our collaboration with AWS allows us to better serve our customers with high-resolution weather prediction systems that feed analytics engines. We’re very excited to see the price/performance of Hpc6a and we expect this to be our go-to Amazon EC2 instance choice for HPC workloads going forward”
Lars Ewe, Chief Technology Officer, DTN

“CFS has benefited greatly from high performance computing. The new Amazon EC2 Hpc6a has been a game changer in many respects. We’ve been able to increase the speed of many simulation tasks, cut runtimes approximately in half, and reduce our computing costs by over 50%. As CFS works to bring clean, limitless commercial fusion energy to the grid, we’re excited to work with AWS and their HPC team.”
Nate O'Farrell, Head of IT Infrastructure, Commonwealth Fusion Systems

“With Amazon EC2 Hpc6a instances, Metro Weather is able to achieve up to a 43% performance improvement over C5n instances for its wind forecasting workloads. Additionally, this performance boost over C5n instances required 50% less compute instances and reduced compute costs by as much as 75%. Hpc6a’s impressive price/performance benefits make it an ideal choice to power Metro Weather’s wind forecasting HPC platform.”
Jun-ichi Furumoto, CEO, Metro Weather Co., Ltd.
Product Details
The Hpc6a instance features the 3rd generation AMD EPYC 7003 series processors with up to 3.6 GHz all-core turbo frequency built on a 7nm process node for increased efficiency.
Instance Size |
Physical Cores |
Memory (GiB) |
EFA Network Bandwidth (Gbps) |
Network Bandwidth (Gbps)* |
Hpc6a.48xlarge |
96 |
384 |
100 |
25 |
Get started with AWS

Start building in the console
Begin building with step-by-step guides to help you launch your AWS project.