Summit

About OLCF
- - ORNL’s supercomputing program grew from humble beginnings to deliver the most powerful system ever seen. On the way, it has helped researchers deliver practical breakthroughs and new scientific knowledge in climate, materials, nuclear science, and a wide range of other disciplines.
- Close
OLCF Resources
- - Learn more about the OLCF’s comprehensive suite of hardware and software resources for the creation, manipulation, and retention of scientific data. For User support in using these resources, please visit the For Users section of this website.
- - Compute Systems
- - Data & Visualization Resources
- Close
R&D Activities
- - Learn more about the OLCF’s comprehensive suite of hardware and software resources for the creation, manipulation, and retention of scientific data. Please note that the For Users section of this website provides extensive information on accessing and employing these resources.
- Close
Science at OLCF
- - The Oak Ridge Leadership Computing Facility (OLCF) engages a world-class team from national laboratories, research institutions, computing centers, universities, and vendors to take a dramatic step forward to field a new capability for high-end science.
- Close
For Users
- - Contact Support
    
    Need assistance from a trained OLCF support staff member? We're here to help.
    
    User Assistance Center
    
    Submit a Support Ticket
    
    Call: 865.241.6536
    
    Email: help@olcf.ornl.gov
    
    Status Tweets: @olcfstatus
- - OLCF User Docs
- - Training
- - Center Announcements
  - Acknowledgement Statement
- Close
OLCF Media
- - Look here for the latest news, reports, and graphics from the OLCF, including branding tools, logos, and acknowledgement statements. You will also find archived information and resources like annual reports and operational assessments.
- Close

About OLCF
- - ORNL’s supercomputing program grew from humble beginnings to deliver the most powerful system ever seen. On the way, it has helped researchers deliver practical breakthroughs and new scientific knowledge in climate, materials, nuclear science, and a wide range of other disciplines.
- Close
OLCF Resources
- - Learn more about the OLCF’s comprehensive suite of hardware and software resources for the creation, manipulation, and retention of scientific data. For User support in using these resources, please visit the For Users section of this website.
- - Compute Systems
- - Data & Visualization Resources
- Close
R&D Activities
- - Learn more about the OLCF’s comprehensive suite of hardware and software resources for the creation, manipulation, and retention of scientific data. Please note that the For Users section of this website provides extensive information on accessing and employing these resources.
- Close
Science at OLCF
- - The Oak Ridge Leadership Computing Facility (OLCF) engages a world-class team from national laboratories, research institutions, computing centers, universities, and vendors to take a dramatic step forward to field a new capability for high-end science.
- Close
For Users
- - Contact Support
    
    Need assistance from a trained OLCF support staff member? We're here to help.
    
    User Assistance Center
    
    Submit a Support Ticket
    
    Call: 865.241.6536
    
    Email: help@olcf.ornl.gov
    
    Status Tweets: @olcfstatus
- - OLCF User Docs
- - Training
- - Center Announcements
  - Acknowledgement Statement
- Close
OLCF Media
- - Look here for the latest news, reports, and graphics from the OLCF, including branding tools, logos, and acknowledgement statements. You will also find archived information and resources like annual reports and operational assessments.
- Close

Introducing Summit

The biggest problems in science require supercomputers of unprecedented capability. That’s why the US Department of Energy’s Oak Ridge National Laboratory (ORNL) launched Summit, a system 8 times more powerful than ORNL’s previous top-ranked system Titan. Summit is providing scientists with incredible computing power to solve challenges in energy, artificial intelligence, human health, and other research areas, that were simply out of reach until now. These discoveries will help shape our understanding of the universe, bolster US economic competitiveness, and contribute to a better future.

Read Press Release Exaop Achievement

Scientific Impact

Dan Jacobson

Deep Learning for Human Systems Biology

Your team is using deep learning to mine experimental and computational data for new insights into biological systems. Can you tell us more about that?
We're trying to understand all the molecular interactions happening in a human cell that lead to the complex traits of an individual and whether those traits result in diseases like Alzheimer's or conditions like chronic pain that can lead to opioid addiction.

For opioid addiction studies, we are collaborating with the US Department of Veterans Affairs to identify patients who have been exposed to opioids and either developed addiction or did not. By combining clinical records and the underlying genomic information for these patients, we can better understand the complex genetic architecture responsible for addiction.
Summit may be considered the world's "smartest" supercomputer for solving problems using artificial intelligence, including deep learning. Why is Summit effective for deep learning projects such as yours?
Summit is a fantastic match for many of our projects. One of the primary reasons for this is Summit's "fat node" architecture, which means it has a lot of memory per node. Another reason is Summit's Volta GPU chips, which are tailor-made for deep learning due to their built-in parallelism.

CS Chang

Plasma Fusion with XGC

What is XGC, and why is it an important application?
We are developing XGC, coupled with other codes, to be a virtual fusion reactor that models the behavior of plasma-the hot gas medium in which particles generate fusion energy. Understanding plasma behavior at a fundamental level is critical for fusion experiments like ITER, which may help meet global energy demand by using seawater to fuel plasma in a fusion device without contributing to the greenhouse effect or producing long-term waste.
What science problem do you hope to solve on Summit?
We plan to better predict the performance of fusion experiments by more realistically simulating particle behavior in a part of the plasma known as the edge that interacts with the interior of the device, or tokamak. Simulation of the edge part of a fusion reactor requires extreme scale computers due to the complexity of the problem.

Joe Oefelein

Combustion with RAPTOR

What kind of science problem will you solve on Summit?
Our focus is turbulent reacting flows in propulsion and power systems-or how combustion takes place in turbulent environments such as gas turbines or car engines. Our code, RAPTOR, is a fluid dynamics solver that incorporates the multiscale and multiphysics processes at play in these systems.
How will Summit help advance your science?
First, just because of the sheer size of the machine, Summit will allow us to work on problems that we haven't been able to do before. Second, Summit will speed our time-to-solution, enabling us to carry out hundreds of runs with RAPTOR in a short time. This will enhance the information we get from our solutions by facilitating detailed parametric studies across a wide range of operating conditions.

Bronson Messer

Astrophysics with FLASH

You and many of your colleagues have been using supercomputers at Oak Ridge for over 15 years. Can you tell us a little bit about your research team and what you study?
We're computational stellar astrophysicists, which means we're interested in how stars live and die. One of the things we're especially interested in is the elements stars produce when they die. We'd like to quantify how that stardust is made and disseminated in the galaxy.
What do you think you will solve using Summit that you have not been able to solve using previous supercomputers?
We'd like to more accurately model nuclear burning that takes place in a supernova, which requires multiple scales of physics calculations, from large-scale hydrodynamics (fluid motion) to small-scale particle interactions. Summit is really going to let us crank up the realism on those small scales. We expect to increase the number of elements simulated in nuclear burning from about 13, which is the state-of-the-art today, to about 160.

Gina Tourassi

Deep Learning for Cancer Research

Your team at the ORNL Health Data Sciences Institute focuses on combining supercomputing and deep learning to advance cancer treatment planning. How so?
We all know cancer is a major public health concern. My team is collaborating with the National Cancer Institute to support implementation of a more advanced population-based cancer surveillance program. We leverage recent advances in artificial intelligence (AI) to enable automated and accurate capture of important cancer surveillance data elements from clinical text documents. We're essentially training computers with large volumes of data to read medical documents and extract important information. Such information is used to help doctors determine the best treatment for each patient and improve population health outcomes.
Why is Summit suited for training an AI application like this?
Summit offers the fastest compute cores; vast amounts of memory, which is important for data-intensive applications such as ours; and efficient communication among cores. These are very important aspects in the Summit architecture for the problem we're solving.

Paul Kent

Materials with QMCPACK

Your team studies the properties of materials using quantum mechanics equations, or physics at the level of individual atoms and electrons. What is a material that you are excited to study on Summit?
Summit will let us study many new materials, but one famous class of materials that we're interested in simulating is high-temperature superconductors, which conduct energy without any loss of efficiency. It would be a breakthrough if we could explain those.
Your team has worked on a new algorithm for QMCPACK that has increased its speed and scale. What aspect of Summit's architecture will help boost QMCPACK performance?
For us, the large on-node memory is very important for increasing the range and complexity of materials and physical phenomena that we can study.

A Supercomputer Suited for AI

Memory Where It Matters

Like real estate, the value of a supercomputer’s memory is closely tied to location. Summit’s sizable local memory, including high-bandwidth memory on each GPU, gives AI researchers a convenient launching point for data-intensive tasks. Minimal data movement means researchers can run deep-learning networks faster and achieve greater accuracy.

GPU Brawn

Training AI algorithms requires processors that can handle a mathematical workout. Each of Summit’s 4,608 nodes contains six deep-learning–optimized GPUs packed with more than 21 billion transistors. And because deep learning requires less precision than traditional scientific computing requires, Summit holds the potential to deliver exascale-level performance for AI algorithms that scale.

CPU Brains

The versatility of Summit’s 22-core Power9 CPUs (two CPUs per node) gives scientists a processor that can manipulate and move data efficiently. The latter task is facilitated by NVLink high-bandwidth technology built into all of Summit’s processors that supplies the next-generation “data superhighways” needed to keep pace with high-speed computation.

Node Key

1. Memory Where It Matters

2. GPU Brawn

3. CPU Brains

What makes a supercomputer smart?

In the case of the Oak Ridge Leadership Computing Facility’s newest leadership-class system, Summit, the answer is in the architecture. Summit, an IBM AC922 system, links more than 27,000 NVIDIA Volta GPUs with more than 9,000 IBM Power9 CPUs to provide unprecedented opportunities for the integration of artificial intelligence (AI) and scientific discovery.

Applying AI techniques like deep learning to automate, accelerate, and drive understanding at supercomputer scales will help scientists achieve breakthroughs in human health, energy, and engineering and answer fundamental questions about the universe.

The arrival of AI supercomputing and Summit means science has never been smarter.

Download AI Infographic PDF

Summit Specs

Application Performance

200 PF

Number of Nodes

4,608

Node performance

42 TF

Memory per Node

512 GB DDR4 + 96 GB HBM2

NV memory per Node

1600 GB

Total System Memory

>10 PB DDR4 + HBM2 + Non-volatile

Processors

2 IBM POWER9™ 9,216 CPUs
6 NVIDIA Volta™ 27,648 GPUs

File System

250 PB, 2.5 TB/s, GPFS™

Power Consumption

13 MW

Interconnect

Mellanox EDR 100G InfiniBand

Operating System

Red Hat Enterprise Linux (RHEL) version 7.4

Summit cabinet dimensions (24" x 76.5" x 49").

Download Specs Infographic PDF

We invite you to read our SC18 paper titled, "The Design, Deployment, and Evaluation of the CORAL Pre-Exascale Systems," for more information about the Summit system.

Citation: Sudharshan S. Vazhkudai, Bronis R. de Supinski, Arthur S. Bland, Al Geist, James Sexton, Jim Kahle, Christopher J. Zimmer, Scott Atchley, Sarp H. Oral, Don E. Maxwell, Veronica G. Vergara Larrea, Adam Bertsch, Robin Goldstone, Wayne Joubert, Chris Chambreau, David Appelhans, Robert Blackmore, Ben Casses, George Chochia, Gene Davison, Matthew A. Ezell, Tom Gooding, Elsa Gonsiorowski, Leopold Grinberg, Bill Hanson, Bill Hartner, Ian Karlin, Matthew L. Leininger, Dustin Leverman, Chris Marroquin, Adam Moody, Martin Ohmacht, Ramesh Pankajakshan, Fernando Pizzano, James H. Rogers, Bryan Rosenburg, Drew Schmidt, Mallikarjun Shankar, Feiyi Wang, Py Watson, Bob Walkup, Lance D. Weems, Junqi Yin, "The Design, Deployment, and Evaluation of the CORAL Pre-Exascale Systems," Proceedings of Supercomputing 2018 (SC18): 31th Int'l Conference on High Performance Computing, Networking, Storage and Analysis, Dallas, TX, November 2018.

By the Numbers

A 200-petaflop machine, Summit can perform 200 quadrillion (peta-) floating point operations per second (-flops). If every person on Earth completed one calculation per second, it would take the world population 305 days to do what Summit can do in 1 second.

At over 340 tons, Summit’s cabinets, file system, and overhead infrastructure weigh more than a large commercial aircraft.

For some AI applications, researchers can use less precise calculations than flops, potentially quadrupling Summit’s performance to exascale levels, or more than a billion billion calculations per second.

Occupying 5,600 sq. ft. of floor space, Summit could fill two tennis courts.

Summit is connected by 185 miles of fiber optic cables—or the distance from Knoxville to Nashville.

Summit’s file system can store 250 petabytes of data, or 74 years of high-definition video.

More than 4,000 gallons of water pump through Summit's cooling system every minute, carrying away about 13 megawatts of heat.

Faces of Summit

Gallery

Top

Contact Support

Contact Support

Introducing Summit