THE POWER AND POTENTIAL OF EXASCALE SUPERCOMPUTING FROM HPE
HPE is delivering Frontier to Oak Ridge National Lab, a system that will accelerate innovation in science and technology and maintain US leadership in high-performance computing and artificial intelligence.
Accelerating scientific discovery
From pharmaceutical research, to extreme weather prediction, to space exploration, high performance computing systems speed discovery, increase accuracy, and stretch the limits of human imagination as we explore new frontiers.
Computing at scale with flexibility
With the rise of cloud computing and the availability of HPC as a service, companies have the opportunity to tackle projects that require heavy duty compute on an on-demand basis. This flexibility enables unprecedented project speed, agility, and scalability.
Powering the next generation of artificial intelligence
AI and machine learning are moving from labs to industry, and supercomputers provide critical infrastructure for intensive data processing. With businesses across many sectors facing the increasing need to analyze and gain insights from large data sets, supercomputing will become the ultimate competitive advantage.
Today’s science, technology, and big data questions are bigger, more complex, and more urgent than ever. Answering those questions demands and entirely new approach to computing.
Today’s science, technology, and big data questions are bigger, more complex, and more urgent than ever. Answering those questions demands and entirely new approach to computing.
Answer your questions today and be ready for what comes next with the most comprehensive supercomputing technology portfolio available for any data center.
Answer diverse and data-intensive questions with an interconnect purpose-built for flexible, heterogenous compute technology.
Create the next generation of applications for innovation and discovery with the industry’s most comprehensive, flexible, and validated HPC and AI software stack.
HPE’s high performance computing storage is as well suited for the unique storage demands of AI as traditional all-flash enterprise file storage and is as scalable and cost effective as traditional parallel storage.
HPE SUPERCOMPUTING
A standard 19" rack configuration supercomputer with the flexibility for smaller and also air-cooled solutions. The current compute platform is the Apollo 2000 Gen10 Plus with 4 AMD® EPYC® 7002 nodes in a 2U compute server with HPE Cray software and HPE Slingshot networking installed.
For highest density and efficiency, liquid-cooled cabinetry supports direct liquid cooling of all components in a highly dense bladed configuration. These cabinets contain eight chassis, each of which supports up to eight compute and eight switchblades. The orthogonal orientation of compute to switchblades enables a cable less connection.
NCAR Advancements Equip First Responders With More Useful Information, Faster
The Institute of Statistical Mathematics refurbished its statistical science supercomputer system to achieve parallel computing performance that was over four times more powerful than its legacy system.
Challenge
Empower emergency responders with fast, effective, granular predictive models of wildfire behaviour.
Solution
Build more sophisticated models by simulating how weather and fire interact, combining machine learning with physical data – without sacrificing modelling speed.
“The resolution of weather forecasts today is a few miles. For firefighters to understand what wildfires might do and how to respond, they need models that are significantly finer. Upgrading to the latest generation of supercomputer will give us the processing power we need to enable those kinds of simulations.”
Result
- Support insights into terrain, conditions to within 30 metres
- Equip stakeholders to get predictions within 1–3 hours
- Improve ability to model variables and quantify uncertainties