During AWS re:Invent 2019, we announced a number of High Performance Computing (HPC) innovations including the Amazon EC2 M6g, C6g, and R6g instances powered by next-generation Arm-based AWS Graviton2 Processors. We also recently announced that new AMD-powered, compute-optimized EC2 instances are in the works.
Today, I'm happy to share some exciting news about our HPC solutions. On November 18, AWS won six HPCwire Readers' and Editors' Choice Awards at SC19, the International Conference for High Performance Computing, Networking, Storage, and Analysis.
Each year, these awards are determined by the global HPC community to recognize the most outstanding individuals, organizations, products, and technologies in the industry. HPCwire is the top news and information resource covering the fastest computers in the world—and the people who run them—since 1987.
There were five categories where AWS was recognized for its industry leadership in HPC.
Best HPC Cloud Platform
AWS won Best HPC Cloud Platform for the second year in a row. With innovative technologies like EFA and the AWS Nitro System, you can achieve high performance without compromising on scale and agility. EFA provides scalable, high-bandwidth, low-latency, interinstance communications. AWS Nitro System almost entirely eliminates the virtualization penalty by delivering practically all of the resources of a server to the instances.
Top 5 New Products or Technologies to Watch
Elastic Fabric Adapter (EFA), a network interface for EC2 instances, was recognized in Top 5 New Products or Technologies to Watch. EFA lets you run applications that need high levels of internode communications at scale on AWS. With EFA, HPC applications using the Message Passing Interface (MPI) and ML applications using NVIDIA Collective Communications Library (NCCL) can scale to thousands of CPUs or GPUs. EFA is enabled on a range of Nitro-based instances including Amazon EC2 C5n, P3dn, I3en, R5n, M5n, G4dn, R5dn, and M5dn instances.
Best Use of HPC in the Cloud (Use Case)
AWS won Best Use of HPC in the Cloud for two use cases: Astera Labs and Ocean Conservancy.
Astera Labs, working with APN Premier Consulting Partner Six Nines, used an electronic design automation (EDA) workflow to create the industry's first Peripheral Component Interface Express (PCIe) v5.0 retimer. This new semiconductor chip was designed entirely in the cloud using Amazon EC2 C5, z1d, and R5 instances to simulate workloads.
Ocean Conservancy used AWS HPC solutions to perform over 75 50-year ocean simulations to understand stressors on the ocean. The stressors lead to the intricate, and potentially catastrophic, effects that climate change is having on our underwater ecosystems. The computational density of this would never have been possible on premises, even using the supercomputers available at the world's largest universities.
Best Use of HPC in Manufacturing
AWS won Best Use of HPC in Manufacturing for Western Digital, an AWS customer. Western Digital built an HPC cluster and used it to simulate crucial elements of upcoming head designs for their next-generation hard disk drives (HDD). This simulation uses Spot Instances to perform 2.5 million tasks, using 1 million vCPUs, in only eight hours—down from 20 calendar days.
Best Use of HPC in Financial Services
AWS won Best Use of HPC in Financial Services with AWS customer, W.R. Hambrecht. The venture firm developed and refined a machine learning (ML)–investment assessment system, the Market Exaptation Simulation Engine (MESE). The big data platform runs on AWS and, because of it, W.R. Hambrecht stated that it has improved the company's rate of picking successful startups by 3x.
With our HPC solutions, we built an integrated suite of services that enable you to quickly and easily build and manage HPC clusters in the cloud, running the most compute-intensive workloads. And we're continually innovating our compute, storage, and networking services to give you the most elastic and scalable cloud infrastructure for your HPC applications.
For example, we have the most powerful GPU instances for ML and graphics workloads. We also offer the fastest processor in the cloud (with EC2 z1d, sustained all core turbo clock speed of 4.0 GHz) to support single-thread performance for compute-intensive workloads like gaming and exploratory data analysis (EDA). And we're the only cloud provider to offer 100-Gbps Ethernet connectivity.
With virtually unlimited capacity, engineers, researchers, and HPC system owners can innovate beyond the limitations of on-premises HPC infrastructure. And they also have instant access to services like data analytics, artificial intelligence (AI), and ML, so that they can transform traditional HPC workflows and get results faster.
We're excited to see broad adoption of our HPC solutions across every industry vertical. Today, more cloud-based HPC applications run on AWS than on any other cloud. These workloads span the traditional HPC applications, like genomics, computational chemistry, financial risk modeling, computer-aided engineering, weather prediction, and seismic imaging. They also cover emerging applications, like ML, deep learning, and autonomous driving. Customers like Bristol-Myers Squibb, FINRA, BP, and Autodesk are using AWS to run their most critical HPC workloads.
To learn more about how AWS is reinventing HPC infrastructure, watch the Monday Night Live keynote with Peter Desantis, AWS VP of Global Infrastructure and Customer Support, from earlier this week at AWS re:Invent.