New HPE+AMD Server Solution Enables Fast, Accurate Data Exploration
Dedicated to advancing scientific discovery, the University of Notre Dame’s Center for Research Computing (CRC) enables multidisciplinary research through advanced computation, data analysis, and other digital tools. When the CRC decided to add a new high performance computing (HPC) cluster, one of the partners they turned to was AMD—a provider of technologies used to help solve some of the world’s toughest challenges.
According to the CRC’s Associate Director, Professor Paul Brenner, “The Center for Research Computing helps faculty pursue big ideas and discoveries through larger, enterprise-scale infrastructure.”
The CRC’s HPC facilities provide more than 30,000 cores of computational power (roughly the equivalent compute capacity of 7,500 laptops) to train next-generation scientists and engineers to tackle some of the world’s most critical applications—including cancer, environmental change, ecosystem modeling, and global economics.
Challenges: Increasing Demands for Performance and Bandwidth
Much of the current research being done at CRC requires both high performance processing and high memory bandwidth to run custom codes with infinite needs for speed while accessing massive data sets. For example, studying the complexities of air pollution requires advanced data modeling, which can in turn help health professionals mitigate human cardiovascular and respiratory disease.
Solution: AMD-based HPC Cluster from HPE
To update its systems, the CRC turned to an AMD and HPE solution of HPE ProLiant DL385 Gen10 servers running AMD EPYC™ 7000 Series processors. The AMD EPYC™ processor family offers the flexibility to choose from 8 to 32 cores, 64 threads, and 8 memory channels with up to 2TB of memory per socket for high-capacity processing with outstanding memory throughput.
Importantly, the AMD EPYC™ processor-based HPE ProLiant DL385 Gen10 servers give the CRC the memory capacity, bandwidth, and processor cores to cost-effectively and efficiently run memory-intensive workloads. They also enable its researchers to analyze large data sets quickly, helping them solve complex problems in an accelerated manner.
As Brenner explains, “If you don’t have good bandwidth to memory, you’ve got a bottleneck. The AMD processor provides high-bandwidth access between the many-core CPU chipset and the RAM which effectively facilitates the larger, high-resolution models that our faculty are interested in.”
“It provides us with additional capabilities that we didn't have before, with systems that yield high performance via tightly coupled compute cores and large bandwidth high-speed RAM,” says Brenner. “This is a major architectural upgrade and change, across the board.”
Application: Studying the Impact of Air Pollutants on Human Health
Paola Crippa, Assistant Professor of Civil and Environmental Engineering and Earth Sciences uses the CRC cluster to understand and model air pollution phenomena that can negatively impact human health—such as increased risks of cardiovascular and respiratory diseases. “The atmosphere is a very complex system which continuously evolves with multiple dynamic and chemical processes. Therefore, to understand the fundamental mechanisms of air pollution we need sophisticated numerical models to account for all these processes, which are very expensive to run,” says Crippa.
Some of Professor Crippa’s model simulations focus on wildfire events in Southeast Asia. “One of the key aims of my research is to use the output of these complex atmospheric models to quantify societal impacts of particles, as well as their effect on regional and global climates,” explains Crippa. “My research has benefited from the use of these state-of-the-art computational resources since they allow me to run simulations at very high resolution.”
Application: Understanding the Carbon Connection between Climate and Plants
David Medvigy, Associate Professor of Biological Sciences, studies the relationship between deforestation and the atmosphere. His research depends upon computation of vast amounts of data to show that the size of deforested areas has a large impact on results—one square mile may make things wetter, but a vaster cut area may make things drier.
“The CRC provides integral resources to my group. We run primarily two different numerical models—both are about 100,000 lines of code,” reveals Medvigy. “We use a lot of information that we get from field observation in our modeling work,” he adds, such as temperature, incoming radiation, precipitation, and soil conditions.
“With increasing amounts of data and more sophisticated understanding of processes that are actually occurring in nature, we can write more sophisticated codes. And to be able to run those, it’s critical to be able to have the right computational resources,” says Medvigy. “Speed is indeed very important. The faster the machines are, the more simulations that I can do.”
Results: A Partnership for Faster, More Detailed Problem-Solving
The kinds of innovative research occurring at Notre Dame are pushing the limits of what’s possible from HPC systems serving the greater good. “This kind of research can certainly have a direct influence on society’s behavior and the policies we make regionally, nationally, and globally. It can help us make better decisions for safer and more sustainable communities,” states Brenner.
As one of the first major rollouts of an AMD EPYC™ processor-based cluster in North America, the CRC installation has been a success. “We got it up and running on-schedule. It ran all of our codes out of the box, as expected. AMD continues to work with us to take our applications to greater scale and higher performance,” says Brenner.
AMD Account Executive, Dan Marowitz adds, “AMD takes a great deal of pride in helping Notre Dame scientists and engineers conduct important research that may improve people’s lives or the state of our planet. We look forward to continuing our support and working with the CRC to expand the cluster's capacity as their needs evolve.”
ABOUT THE CENTER FOR RESEARCH COMPUTING
The Center for Research Computing at the University of Notre Dame is an innovative and multidisciplinary research environment that supports collaboration to facilitate discoveries in science and engineering, the arts, humanities and social sciences, through advanced computation, data analysis and other digital research tools. The Center enhances the University’s cyberinfrastructure, provides support for interdisciplinary research and education, and conducts computational research.
For more than 45 years AMD has driven innovation in high performance computing, graphics, and visualization technologies— the building blocks for gaming, immersive platforms, and the datacenter. Hundreds of millions of consumers, leading Fortune 500 businesses, and cutting-edge scientific research facilities around the world rely on AMD technology daily to improve how they live, work, and play. AMD employees around the world are focused on building great products that push the boundaries of what is possible. For more information about how AMD is enabling today and inspiring tomorrow, visit amd.com/epyc and amd.com/corporate-responsibility.
ABOUT HEWLETT PACKARD ENTERPRISE
We are in the acceleration business. We help customers use technology to slash the time it takes to turn ideas into value. In turn, they transform industries, markets and lives. Some of our customers run traditional IT environments. Most are transitioning to a secure, cloud-enabled, mobile-friendly infrastructure. Many rely on a combination of both. Wherever they are in that journey, we provide the technology and solutions to help them succeed.
© 2018 Center for Research Computing Images and video courtesy of the University of Notre Dame
© Copyright 2018 Hewlett Packard Enterprise Development LP. HPE and ProLiant are trademarks of Hewlett Packard Enterprise
© 2018 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo, Epyc and combinations thereof are trademarks of Advanced Micro Devices, Inc. Other product names used in this publication are for identification purposes only and may be trademarks of their respective companies.