HPC Workloads in the Cloud, Powered by AWS and Intel

At a Glance

  • Your customers can choose from an array of Amazon Elastic Compute Cloud (Amazon EC2) instances built on Intel® Xeon® Scalable processors and designed to deliver performance to suit your high-performance computing (HPC) workload needs.

  • Saildrone completes precise environmental data collection at scale with AWS EC2 C5n instances and the powerful Intel® Xeon® Platinum 8000 series processor.

  • Astera Labs has deployed thousands of Amazon EC2 R5, C5 and z1d instances to run more than 1,000 chip simulations, reducing product development time and costs.


“AWS and Intel have a 14+ year engineering and business partnership focused on delivering cutting edge technology with a focus on value for our customers’ unique needs. Together, we tailored the Amazon EC2 c5n and z1d instances to not only take full advantage of Intel® Xeon® Scalable processor technology but also to speed up inter-node communication and to maximize single-threaded performance.”

Rebecca Weekly, senior director of business strategy and platform enabling, senior principal engineer Intel Corporation.

The collaboration between AWS and Intel is expanding the limits of what customers can achieve with HPC workloads in the cloud. Using feedback from customers on the cost, security and performance constraints they faced on-premises, AWS and Intel co-engineered a new instance with customization at its heart. As a result, with Amazon EC2 c5n and z1d instances optimized for HPC and based on custom Intel® Xeon® Scalable processors, your customers can now migrate HPC workloads to the cloud with even greater confidence. At re:Invent 2020, AWS and Intel announced new capabilities for HPC workloads for financial services, gaming and simulation modeling for high-performance, high-frequency Amazon EC2 M5zn instance, based on 2nd generation Intel® Xeon® Scalable processors.

Amazon EC2 z1d high-frequency instances push the performance and power boundaries of Intel® Xeon® Scalable processors, providing high single-thread performance with the highest sustained all-core frequency of any cloud instance.1 With both high compute performance and high memory, z1d is well-suited to HPC workloads with high per-core licensing costs. Intel® Turbo Boost Technology also enables the processor to achieve a higher clock rate based on the available power and thermal headroom. There are seven different z1d instance sizes, including a bare metal instance, providing up to 48 vCPUs per instance.1

Put simply, HPC on AWS offers an elastic, scalable, and customizable computing environment capable of handling a broad range of high-performance use cases. Customers can also improve productivity with AWS’ per second billing for compute resources, ensuring users of Amazon EC2 instances can quickly scale up and down as needed. With high availability in multiple locations worldwide, it’s possible to protect against failures and ensure sufficient capacity for even the most compute-intensive workloads.2 Customers can also get started quickly, securely, and cost effectively, with the virtually unlimited infrastructure capacity necessary for engineers and researchers to innovate without constraint.

Customers in market today are already experiencing the benefits of AWS and Intel’s 10+ year engineering relationship and shared passion for innovation. Explore the stories below to see for yourself why – and how – AWS and Intel are ‘better together’.

Accelerated Chip Development with Rapid Simulations3

Semiconductor company Astera Labs designs purpose-built connectivity solutions for data-centric systems. Getting to the final, critical tape-out phase of development means quickly performing complex simulations powered by enormous compute capacity. If the company can’t complete enough simulations during its chip design and development process, potential flaws could be missed. Astera needed a solution that enabled it to meet tape-out deadlines without compromising on quality.3

Astera Labs built an HPC environment running thousands of Amazon EC2 R5, C5 and z1d instances powered by Intel® Xeon® Scalable processors. The z1d instances are used by Astera Labs for long-running simulations that require the fastest-possible virtual machines. That’s because, with custom Intel® Xeon® Scalable processors, the z1d instances are among the fastest in the cloud. They provide high single-thread performance and sustained core frequency of up to 4.0 GHz.

Astera Labs now has the compute resources to run more than 1,000 simulations overnight and was able to complete the tape-out of a new chip that meets industry standards for high bandwidth and low-latency interconnects. The company has also saved several months of time that would otherwise have been spent correcting mistakes, while saving millions of dollars overhauling a new chip design following a tape-out.

To find out more about how Astera Labs is benefiting from the collaboration between AWS and Intel, take a look at this video.

Uninterrupted Environmental Data Collection4

Saildrone designs, manufactures and operates a global fleet of wind and solar-powered ocean drones that monitor the planet in real time. Each Saildrone is designed to carry sensors that measure atmospheric and oceanographic variables with high precision in often challenging conditions. They are powerhouses of data collection, taking measurements including wind speed/direction, barometric pressure, water temperature, atmospheric carbon dioxide, and many other metrics.4

One of the ways Saildrone uses this data is to estimate the biomass of fish stocks via sonar, ultimately developing statistical biomass models over vast swathes of the planet. The faster Saildrone counts the migrating fish, the more accurate the model will be. But Saildrone determined that building the dedicated compute and infrastructure resources required to manage this modeling wasn’t a viable use of its resources. As a result, Saildrone opted for Amazon EC2 C5n instances to provide the scale and resilience it needs to process and monitor the data it collects. The powerful Intel® Xeon® Platinum 8000 series processor in C5n instances delivers the performance Saildrone needs for its large set of data points.

Saildrone’s next frontier is to map the entire global seabed, 85 percent of which is currently unexplored. Its goal is to complete the first map by 2030.

Customers who choose to deploy AWS C5n.18xlarge instances can also benefit from added support for AWS Elastic Fabric Adapter. The combination of AWS Elastic Fabric Adapter and its support by the Intel® MPI Library, a multi-fabric message-passing library based on an open industry initiative, improves the efficiency of inter-node communication—and improves performance for scaling HPC workloads.

Mastering Scalable, Highly Precise Search Algorithms5

The Descartes Labs Platform aims to give data scientists the best geospatial data and modeling tools possible to scale computer vision-based applications. One challenge the company examined was how to identify large oil storage tanks located across the surface of the Earth. Identifying every single tank required a refined, careful algorithmic search capability.

To address this, Descartes Labs built a computer vision-based solution using Amazon EC2 C5 instances and Intel® Xeon® Platinum processors. Descartes Labs also took advantage of Intel® Advanced Vector Extensions 512 (Intel® AVX-512). With explicit vector instructions, the company’s object finder tool delivers results at high speed. Descartes Labs found that code executes at more than 50 percent peak FLOPS, producing higher precision and recall than off-the-shelf solutions.5

Scalable Turbine Modeling to Meet Customer Demand6

Avio Aero, a GE Aviation business, uses an application called TRAF to run the computational fluid dynamics (CFD) simulations required to develop the latest generation of aircraft turbines. However, the application is resource-intensive, meaning meeting the needs of a single customer could require 100 percent of the company’s on-premises capacity for two to three months.6 Avio Aero built an HPC solution using Amazon EC2 C5 instances based on Intel® Xeon® Scalable processors. This solution meets demand that cannot be handled on-premises. AWS computing nodes are deployed in placement groups to reduce network latency and improve performance.

Taking advantage of AWS’s flexible scalability and a pay-as-you-go price structure, Avio Aero ensures it only pays for the capacity it needs. Runtimes are now three times faster,6 meaning that customer requests can be immediately satisfied, and the company can focus on its continued innovations.

The best evidence of the strong foundation for HPC workloads that Intel and AWS provide together is the success of our customers. Get in touch today to identify customers and proofs of concept that can take advantage of Amazon EC2 r5, c5/c5n and z1d instances powered by Intel.

Stay Up to Date

Sign up to receive news about Intel® technologies for AWS.

By submitting this form, you are confirming you are an adult 18 years or older and you agree to share your personal information with Intel to use for this business request. Intel's web sites and communications are subject to our Privacy Notice and Terms of Use.
By submitting this form, you are confirming you are an adult 18 years or older and you agree to share your personal information with Intel to use for this business request. You also agree to subscribe to stay connected to the latest Intel technologies and industry trends by email and telephone. You may unsubscribe at any time. Intel's web sites and communications are subject to our Privacy Notice and Terms of Use.

Notices and Disclaimers

Software and workloads used in performance tests may have been optimized for performance only on Intel® microprocessors.

Performance tests, such as SYSmark* and MobileMark*, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information visit www.intel.com/benchmarks.

Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details. No product or component can be absolutely secure.

Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel® microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel® microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice.

Refer to https://software.intel.com/articles/optimization-notice for more information regarding performance and optimization choices in Intel® software products.

Intel does not control or audit third-party data. You should consult other sources to evaluate accuracy.

Intel® technologies may require enabled hardware, software or service activation.

Your costs and results may vary.

© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.