Exascale supercomputing is here and it will change the world

OCTOBER 18, 2022 • BLOG POST • JUSTIN HOTARD, EVP AND GENERAL MANAGER, HPC, AI AND LABS AT HEWLETT PACKARD ENTERPRISE

TAP IMAGE TO ZOOM IN

IN THIS ARTICLE

  • Annual Exascale Day celebrates massive achievement and community behind realizing exascale computing
  • Exascale was ushered in by Frontier, the world’s first and fastest exascale supercomputer for the U.S. Department of Energy’s Oak Ridge National Laboratory
  • Exascale will enable a new era of breakthroughs in scientific discovery, artificial intelligence, and early testing for quantum applications 

HPE celebrates Exascale Day with the debut of the Frontier supercomputer, its initial scientific and performance breakthroughs, and the new era it has ushered in 

Supercomputing is vital to global innovation, industry competitiveness, and economic growth.

Nations and organizations rely on supercomputing to solve the world’s toughest problems and realize tremendous societal, environmental, and economic impacts. The massive, specialized computing capabilities of supercomputing are also critical to enabling AI-at-scale, which requires processing, training, and storing large data sets. This capability allows us to build bigger, robust machine learning models needed to accurately predict outcomes and help us make better decisions faster.

And a significant technological leap in supercomputing only makes things better.

That is why on this October 18, we are celebrating a massive achievement in supercomputing – the equivalent to landing on the Moon – of breaking the exascale speed barrier to solve problems up to 10X faster, at 1,000,000,000,000,000,000 calculations per second. Or 1018.

VIDEO: Frontier: The World's First Exascale Supercomputer Has Arrived

Today, we celebrate Exascale Day.

When imagining the benefits of exascale, Dr. Niven Narain, CEO and co-founder of Berg Health, an AI-powered biotechnology company, explained that exascale supercomputing isn’t just about speed for speed sake, but “speed for human sake.”

I couldn’t agree more.

Exascale speed means overcoming previous performance hurdles to unlock scientific insights never possible before, from modeling and simulating interactions more accurately down to the physical, biological, and chemical molecular level, to innovate and solve problems faster.

It is speed that will help advance our understanding of a global virus and accelerate the discovery of new vaccines to combat it. Speed that will accurately predict a hurricane’s landfall to allow enough time to safely evacuate people and save more lives. Speed that will help find ways to create a sustainable future by efficiently harnessing more clean energy from the sun, wind, water, and perhaps soon, nuclear fusion, to power our homes and neighborhoods. 

Realizing the benefits of exascale with Frontier, the world’s first and fastest exascale supercomputer

The speed and capabilities of exascale, ushered in this year with Frontier, a supercomputer built by HPE for the U.S. Department of Energy’s Oak Ridge National Laboratory, will enable global researchers, scientists and engineers to answer questions we never knew to ask.

As a supercomputer dedicated to open science, Frontier will also benefit global communities to collaborate on research and take advantage of the resources, including the ability to access larger AI models that be used predict outcomes faster in scientific discovery. 


VIDEO:
Frontier: Moving the future forward

Using Frontier, scientists and researchers have already achieved initial performance and scientific breakthroughs. These include understanding why certain COVID-19 strains are more transmissible than others, to advancing our knowledge in physics to improve plasma accelerators that create more energy to advance radiation technologies commonly used for a range of commercial purposes. 

Example research includes:

 

  • Analyzing all COVID-19 mutations and transmissibility from one week to 24 hours. In a project that focuses on analyzing large quantities of genomic data, Frontier generated analysis on all COVID-19 genome types to understand how certain variations made the virus more transmissible.

    The analysis revealed there were tens of thousands of variations that made particular strains more prominent in certain parts of the world. Scientists also sped up discovery by 6.3X, from a week to just a day.1

 

  • Speeding up the nuclear fission process by advancing the design of laser-based electron accelerators. By using Frontier, in combination with other top supercomputers, scientists were able to simulate plasma accelerators 500X faster since this effort was started six years ago. Plasma accelerators are used to accelerate charged particles such as electrons, positrons, and ions, to speed up the process of splitting atoms, also known as nuclear fission.

    Particle accelerators today, while less powerful, have become vital in a range of commercial purposes, from treating cancer with radiation therapy to manufacturing semiconductors for computer chips. By advancing plasma simulations to make more powerful particle accelerators, researchers can conduct high-energy physics experiments to accelerate scientific discovery.

 

  • Accelerating patient diagnosis and care by making biomedical literature accessible using graph AI. Researchers used Frontier to conduct an experiment that was never possible before exascale as it relied on large-scale mining, using AI capabilities, of biomedical research literature, which consists of tens of millions of scientific papers. These papers span decades of research in areas such as drug and surgical therapies, diagnostic procedures and preventive measures.

    By applying the computational speed of exascale, researchers constructed graph representations from annotated text databases that lead to better ways to easily search for biomedical information to help improve diagnostics and treatment paths for patients.


Exascale will strengthen the commercial sector to make better, safer and efficient products

Exascale systems like Frontier will also enable commercial organizations, such as 3M, Boeing, General Electric (GE), General Motors, and Procter & Gamble, which already use the U.S. DOE’s supercomputing resources, to advance R&D to design better and safer products that we all use and benefit from each day.

In fact, GE Research, a world-leader with more than a century of expertise and innovation in advancing growth sectors in aviation, energy and healthcare industries, is planning to use Frontier to transform aviation, as part of its “Future of Flight” strategy. GE’s researchers plan to model and simulate computational fluid dynamics, which is core to understanding aircraft system performance, at the exascale level to predict the next generation of jet engine performance, safety, and efficiency without physical testing. This R&D will significantly accelerate product design and development, and speed time-to-market.

This R&D will significantly accelerate product design and development, and speed time-to-market. As part of this research, a major goal of GE’s project is to improve sustainability by designing jet engines that can sufficiently support hydrogen, a zero-CO2-emission fuel, by advancing propulsive efficiency.

Fueling a new era of scientific discovery and engineering with exascale

Given the critical importance that supercomputing plays in underpinning R&D and innovation for a range of sectors that drive economic growth and strengthen national security and defense, the computational power that exascale delivers will further augment efforts required for nations and organizations to stay competitive.

The exascale era will enable engineering breakthroughs that are core to fueling next-generation technologies and opening doorways for innovation. New technology advantages enabled by exascale include:

 

  • Transforming modeling and simulation to gain realistic models – By modeling and simulating at exascale speed and converging those workloads with AI and machine learning capabilities, researchers will gain greater granularity and accuracy in models that are important to unlocking insights. This will have an impact on areas such as accurately modeling the human brain to advance medical research or improving graphics of atmospheric data to better identify weather patterns in a given geological region.

  • Harnessing AI-at-scale to solve the unsolvable – With exascale performance, users can build AI models 4.5X faster and 8X larger, allowing to train more data that can increase predictability and speed time-to-discovery. These models will help advance techniques in areas such as natural language processing and computer vision that are largely used to unlock intelligence.

  • Accelerating the journey to progress quantum technology – Supercomputing continues to be a valuable tool in benchmarking computational advantage. The computing power that exascale delivers will allow researchers to conduct early stages of testing and optimizing for quantum applications to mature quantum technology.
     

Enabling broader enterprise adoption of supercomputing to speed time-to-insight

Bringing technologies, such as purpose-built interconnect, storage, and software in an accessible and easy-to-manage model, can help broader enterprises tackle demanding applications. where we capture the benefits of AI-at-scale to solve problems that were too difficult to characterize using traditional modeling and simulation and the Exascale era gives us the necessary computing power to bridge to early Quantum applications.

These include support to analyze vast amounts of data, or build better machine learning models, to drive outcomes. These capabilities can benefit a range of industries, from speeding up drug discovery for a pharmaceutical company, forecasting critical stock trends and trade, or prevent fraud detection, for a financial services organization, or simulate car crashes for car manufacturers to build and launch safer vehicles.

Our team at HPE is committed to making this a reality by incorporating elements of high-end supercomputing technologies that we used to design and build Frontier, in a smaller footprint or as-a-service through the HPE GreenLake platform.

I am excited to continue sharing updates as we progress toward realizing this vision.

VIDEO: Why is Exascale Monumental?

Recognizing researchers, scientists and engineers who continue to ask "what if, why not, and what's next"

The reality of exascale wouldn’t be possible today if it wasn’t for the researchers, scientists, and engineers who envisioned this for more than a decade. Join us in celebrating Exascale Day, and the people behind this day, with a multi-panel broadcast, starting at 12 p.m. Central Time.

You’ll hear from industry experts with commercial organizations on how supercomputing impacts the products we use every day, and from research scientists across the globe that are committed to making the world we live in safer and more sustainable.

Leaders from the U.S. Department of Energy, such as Dr. Asmeret Asefaw Berhe, director of Office of Science, whom I will have the honor to interview, will discuss the bigger mission for enabling open science with exascale-class supercomputers.

To join, please visit: https://exascale-day.brighttalk.com/

RELATED NEWS