- Amazon EC2›
- Instance types›
- Hpc8a Instances
Amazon EC2 Hpc8a instances
Next-generation high performance computing instances powered by 5th Gen AMD EPYC processors
Why Amazon EC2 Hpc8a instances
Amazon EC2 Hpc8a instances deliver the next generation of high performance computing on AWS, powered by 5th Generation AMD EPYC processors with frequencies up to 4.5 GHz. Designed for tightly coupled HPC workloads, Hpc8a instances provide up to 40% higher performance, 42% greater memory bandwidth, and up to 25% better price performance compared to previous-generation Hpc7a instances. With high core density, industry-leading memory throughput, and low-latency Elastic Fabric Adapter (EFA) networking, Hpc8a instances help engineers, researchers, and scientists accelerate simulations, reduce job runtime, and achieve faster time-to-results.
Benefits
Hpc8a instances deliver significant generational performance improvements for compute-intensive HPC workloads such as computational fluid dynamics (CFD), crash simulations, and high-resolution weather modeling. Higher processor frequency, improved memory bandwidth, and optimized architecture enable faster iteration cycles and shorter simulation runtimes.
With up to 25% better price performance compared to Hpc7a instances, organizations can run more simulations within the same infrastructure budget while maintaining consistent performance at scale.
Support for 300 Gbps Elastic Fabric Adapter (EFA) networking enables low-latency, high-throughput inter-node communication, allowing MPI-based workloads to scale efficiently across large HPC clusters.
Hpc8a instances integrate seamlessly with AWS ParallelCluster, and AWS Parallel Computing Service, enabling customers to deploy and scale HPC clusters quickly while simplifying workload submission and cluster management.
Features
Hpc8a instances use the latest AMD EPYC processors with frequencies up to 4.5 GHz, delivering improved compute throughput and architectural enhancements optimized for both single-threaded and multi-core HPC workloads.
With 42% greater memory bandwidth Hpc8a instances enable faster processing of memory-intensive simulations and numerical workloads.
Hpc8a instances use sixth-generation AWS Nitro cards to offload virtualization, storage, and networking functions to dedicated hardware, delivering near-bare-metal performance and improved security isolation.
Integrated 300 Gbps EFA networking enables low-latency cluster communication for tightly coupled MPI workloads, helping customers scale simulations efficiently across thousands of cores.
Simultaneous multithreading (SMT) is disabled to optimize HPC performance consistency, enabling predictable performance for tightly coupled workloads.
Use cases
Computational fluid dynamics (CFD)
Run aerodynamic and thermal simulations faster using high-frequency processor cores and scalable cluster networking.
Crash and structural simulations
Accelerate automotive and aerospace crash analysis workloads requiring rapid simulation turnaround times.
Weather and climate modeling
Process high-resolution atmospheric simulations within operational time windows using scalable HPC clusters.
Scientific and engineering simulations
Run larger, more complex simulations while reducing overall job completion time.
Instance specifications
|
Instance size
|
Physical cores
|
Memory (GiB)
|
Instance Storage
|
EFA bandwidth (Gbps)
|
Network bandwidth (Gbps)
|
|---|---|---|---|---|---|
|
hpc8a.96xlarge
|
192 |
768 |
EBS only |
300 |
75 |
Hpc8a instances are available in a single 96xlarge size with a 1:4 core-to-memory ratio, enabling customers to right-size compute capacity by customizing the number of cores at instance launch.
Customer Testimonials
Here are some examples of how customers and partners have achieved their business agility, price performance, cost savings, and sustainability goals with Amazon EC2 Hpc8a instances.
Rescale
Rescale is a cloud-based high-performance computing (HPC) platform that enables enterprises to run AI-accelerated modeling and simulation workloads to enable engineering breakthroughs across aerospace, automotive, manufacturing, life science and government sectors.
“The release of Amazon EC2 Hpc8a instances is a significant advancement for Rescale and our customers running demanding HPC workloads. Hpc8a delivers up to 52% speedup on CFD applications and up to 38% speedup on CAE applications (compared to Hpc7a), enabling us to better serve customer requirements for computer-aided engineering simulations with ever-increasing complexity.
Based on 5th Gen AMD EPYC processors with improved network latency and memory bandwidth at the microarchitectural level, Hpc8a instances demonstrate remarkable versatility across Rescale’s diverse customer workloads. From automotive and aerospace to semiconductor design and life sciences, Hpc8a helps customers solve fluid dynamics, structural analysis, and multiphysics simulations while enabling a seamless transition from the current generation and delivering immediate performance gains to engineering teams on the Rescale platform.”
— Radhika Gundavelli, Sr. HPC Engineering Manager, Rescale
Spire
Spire is a global provider of space-based data, analytics, and space services, offering the unique datasets and powerful insights needed for governments and organizations to solve problems on Earth and make better decisions about what to do next in a rapidly changing world. By capturing radio frequency (RF) signals from space, Spire “listens” to the Earth in real-time to reveal the movements, atmospheric patterns, and hidden behaviors that define global operations, safety, and security.
“Powered by data from our proprietary satellite constellation, Spire’s high-resolution weather models are built on a truly differentiated foundation. The transition to Amazon EC2 Hpc8a instances has delivered a 1.5x speedup, representing a generational leap in model performance. That enables us to generate faster, more timely forecasts from complex atmospheric data. For energy, agriculture, and other weather-sensitive industries, that means sharper insight, greater confidence, and better decisions when it matters most.”
— Alex Farr, Team Lead, Software Engineering, Spire Weather & Climate
Weathernews
Weathernews is a global provider of weather intelligence, forecasting, and risk analytics services for enterprise and public sector organizations
"At Weathernews, we provide mission-critical weather information to a wide range of customers, including vessel operators. AWS agility and availability are essential to our business. In our evaluation of Amazon EC2 Hpc8a instances, we observed a 34% performance improvement in our WRF-based production workloads compared to the previous generation. This directly translates into faster, more timely forecasts for our customers. The ability to rapidly adopt the latest CPU technology in the cloud reaffirms why we chose AWS as our platform, and we look forward to further enhancing our services with Hpc8a.“
— Kazunari Takahashi, Forecast Center, Weathernews
Mitsubishi Heavy Industries (MHI)
Mitsubishi Heavy Industries is a global engineering and manufacturing company delivering solutions across energy, aerospace, infrastructure and industrial systems.
“Amazon EC2 Hpc8a instances accelerated our CFD simulations by 36%, redefining the scale and speed at which we can explore advanced steam turbine aerodynamics. This breakthrough enables high-fidelity physics investigations that were previously out of reach and empowers us to pursue more aggressive design innovation on Amazon EC2 Hpc8a.”
—Dr. Soichiro Tabata, Manager, Technical Lead for Steam Turbine and Two-Phase Flow CFD Mitsubishi Heavy Industries”
Getting started
Amazon EC2 Hpc8a instances are available through On-Demand Instances and Savings Plans. Launch instances using the AWS Management Console, AWS CLI, or infrastructure-as-code tools such as AWS CloudFormation and Terraform. Deploy clusters quickly using AWS ParallelCluster or AWS Parallel Computing Service to begin running large-scale simulations in minutes.
Did you find what you were looking for today?
Let us know so we can improve the quality of the content on our pages