AWS Introduces HPC Instances Equipped With AMD’s Latest Server CPUs
Pitching supercomputing power at a reasonable price.
![AMD server CPU AMD server CPU](https://eu-images.contentstack.com/v3/assets/blt8eb3cdfc1fce5194/bltbae35cb6e0514bee/6621136ce09962cbdb8a0e97/720217-AMD-EPYC-bg-angle-1920x600.jpg?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Amazon Web Services has introduced new Elastic Compute Cloud (EC2) instances designed for HPC workloads, powered by AMD’s third generation EPYC processors.
The cloud vendor said the instances codenamed ‘Hpc6a’ offer up to 65 percent better price-performance than similar compute-optimized, x86-based EC2 instances.
Potential applications for all this compute include genomics, computational fluid dynamics, weather forecasting, financial risk modeling, electronic design automation, computer-aided engineering, and seismic imaging.
“Amazon EC2 Hpc6a instances, powered by 3rd Gen AMD EPYC processors, allow organizations the flexibility to run HPC workloads requiring an abundance of compute power, fast memory and storage, and high levels of inter-instance communication without the upfront cost of building and maintaining HPC infrastructure on-premises,” said David Brown, vice president of the EC2 business at AWS.
The instances are available immediately in US East (Ohio) and AWS GovCloud (US-West), with availability in additional AWS Regions planned soon.
These are not the first instances to be equipped with AMD’s wunder-chip; general-purpose EC2 options featuring 3rd gen EPYCs were launched in November 2021.
Real big instances
Each of the new Hpc6a instances is powered by 96 EPYC 7003 series cores clocked at 3.6 GHz, and equipped with 384GB of memory.
They are built on the AWS Nitro System, a combination of dedicated hardware and lightweight hypervisor which claims to deliver practically all of the compute and memory resources of the host hardware to customer instances.
They also have Elastic Fabric Adapter (EFA) enabled by default to provide low latency 100 Gbps bandwidth for inter-node communication within a single Availability Zone. Meanwhile, simultaneous multi-threading is disabled to ensure one physical core per vCPU.
In its announcement, AMD was quick to point out that EPYC processors are used in 73 of the world’s 500 most powerful supercomputers.
“Our processors power all levels of HPC, from exascale systems in research laboratories to flexible HPC cloud computing instances like the new Amazon EC2 Hpc6a instances,” said Dan McNamara, SVP and GM of AMD’s server business.
“AMD EPYC processors provide a powerful solution for Amazon EC2 customers that need access to impressive HPC performance and cloud scalability for their workloads.”
About the Author
You May Also Like