Distinguished Argonne Fellow
Welcome to the First Edition of ECP Update
The Exascale Computing Project (ECP) leadership team will be publishing the ECP Update newsletter as we have announcements, project updates, or items of interest to share. In addition to updates on activities from our four focus areas (Application Development, Hardware Technology, Software Technology, and Exascale Systems), you will see updates on ECP Integration, Project Office, and Industry Council activities and significant meetings and events.
We will also have additional features from our laboratory partners, as well as Industry Council members; and occasionally we will provide links to articles and reports that you might find interesting. Suggestions for the latter are welcome.
The first year of the ECP has been marked by intense activity, as we solicited, selected, and launched efforts in many areas and began the complex but essential task of integrating the projects.
As a recap of recent highlights, we celebrated the project’s one-year anniversary with our first annual meeting of ECP collaborators, attended by more than 450 researchers and collaborating partners in Knoxville, Tennessee, from January 31 to February 2, 2017.
On February 1, we announced the formation of the ECP Industry Council, an external advisory group of executives from some of the nation’s most prominent companies. We held the first council meeting on March 7 and have included a brief update of that meeting in this newsletter.
And on March 2, we announced our fifth co-design center—a partnership among Pacific Northwest, Lawrence Berkeley, and Sandia National Laboratories and Purdue University that will focus on graph analytics.
We now have more than 800 researchers engaged in ECP-funded activities as we move into our second year. The extended ECP community includes 22 US Department of Energy laboratory and collaborating agency partners, 9 private sector companies, 39 university research partners, and 18 members of the Industry Council.
With all the activities that are now under way, there will be a lot of results and insights to track; and we encourage you to follow our progress through the ECP website.
If you are not already a subscriber, be sure to subscribe here.
Director, The Exascale Computing Project
Expert R&D to Build Exascale for Science
The Application Development focus area is responsible for delivering science-based applications able to exploit exascale computation for high-confidence insights and answers to critical problems in national security, energy assurance, economic competitiveness, and healthcare.
“Apps for the nation” are essential for new solutions to national challenges
The United States faces urgent energy, security, and public health challenges that can be solved through science and technology. Many of us download applications (“apps”) on our smart phones or computers to improve our quality of life—to help us be more efficient at our jobs, save money, or increase the security of our homes and personal data. Supercomputing applications can do this on a national scale by enabling us to simulate the insides of nuclear reactors, drug delivery in the body, potential earthquakes and natural disasters, blackouts and electricity demand on the power grid, and other processes that are important to our health, safety, and prosperity.
To solve immensely complex mathematical problems in hours or days rather than years or decades, applications are written in millions of lines of code that must run on thousands of computer processors expertly designed to work synchronously. Good science applications create for us virtual laboratories that take us where we cannot go—inside dangerous, extreme, or unseen processes—and allow us to test many options so that we start with the very best. In this way, good applications save time and lead to new discoveries and better solutions.
With each new generation of supercomputers, application results are faster, more predictive, and more responsive to the pace of industry and the environment. At exascale, (computing at a billion billion calculations per second), applications will improve predictive simulations by moving from model systems that, though complex, make some simplifying assumptions, to system sizes in close agreement with reality—which can increase confidence in decision- and policy-making.
In fall 2016, the ECP’s Application Development focus area initiated 25 projects chartered to address major scientific and technical problems that affect the future of our nation’s scientific leadership, economic competitiveness, and national security. These critical applications are being developed by hundreds of talented scientists and engineers from 45 national research and academic institutions. The first 4 years are focused on the development of models, algorithms, and computational methods; the integration of software and hardware systems; the improvement of exascale system readiness; and the demonstration of each application’s capability to solve the scientific problems for which it was designed. In March 2017, the applications teams completed their first milestone reports and are on track to deliver new science at a new scale.
The Software Technology focus area is developing a comprehensive and coherent software stack that will enable application developers to productively write highly parallel applications that portably target diverse exascale architectures. The software stack spans from low-level operational software to programming models and environments for high-level applications software and includes the software infrastructure to support large-scale data analysis and visualization. To develop this sophisticated software stack, we plan to extend current technologies to exascale where possible, perform R&D required to conceive of new approaches when necessary, coordinate with vendor efforts (i.e., developing software other than that typically developed by vendors), develop common interfaces and services, and build and deploy high-quality and robust software products.
We look forward to bringing you monthly updates from the Software Technology focus area via this monthly newsletter.
The Hardware Technology focus area leverages ECP’s opportunity to have a direct impact on future hardware designs at the component, node, and system architecture levels. Most funding for hardware technology R&D is designated for a handful of vendor-based projects called PathForward, which will support hardware technology designs that meet requirements for ECP technical challenges.
The key ECP technical challenges are massive parallelism, memory and storage, reliability, and energy consumption. To support ECP’s holistic co-design interactions, PathForward will use DOE national laboratory architectural analysis capabilities to determine the best system architecture for DOE’s exascale goals, and abstract machine models, which are theoretical computer hardware and system models.
Exascale Systems focuses on advanced system engineering development by vendors to produce capable exascale systems and test beds for research, co-design activities, and developing and optimizing applications and software.
Looking forward to early access on next-generation systems
The ECP application and software development teams will receive access to early production systems at the Argonne and Oak Ridge Leadership Computing Facilities (ALCF and OLCF, respectively). ALCF hosts the Theta early production system for helping users transition their applications to the new Intel® Xeon PHI™ architecture. The OLCF recently installed a test and development system called Summitdev that is one generation removed from the facility’s next hybrid GPU–CPU computer, Summit; Summitdev will enable developers to test IBM’s POWER processors and NVIDIA’s Pascal GPU architecture before Summit arrives in 2018.
Additionally, ECP users will have limited access to NERSC’s Cori system and the OLCF Titan system through June 30, 2017.
Integration Will Be Critical to Exascale Project Success
The ECP is a holistic approach to reaching capable exascale computing, employing efforts in four key focus areas: Application Development, Software Technology, Hardware Technology, and Exascale Systems. However, just as important as their individual efforts is the integration of these elements into a seamless development environment.
Applications projects will use and regularly assess the software tools being developed by ECP-funded teams, partnering with computer scientists to define and refine specific capabilities. Co-design centers will focus on key collaborative elements and drive additional activities between focus areas. While the Department of Energy facilities will ultimately acquire the next-generation systems, insight and input into future architectures will be funneled through a conduit of hardware technology projects and technical representatives to system vendors. This arrangement fosters a co-design environment in which system design is influenced by application and software needs and, reciprocally, computational and computer scientists can make more informed research and development decisions based on understanding future system design paths.
These integration points and more were discussed during the November 2016 ECP Principal Investigator and Integration Meeting and, more recently, at the January 2017 ECP Annual Meeting, resulting in hundreds of potential contacts between projects. Many of these are currently being documented as integration milestones, while others will evolve over the next year. The success of the ECP depends on more than the sum of its parts. For the United States to realize an effective exascale computing environment, each participant in the ECP must look beyond its own project and see how it can plug in to other focus areas. In this way, productive scientific research will be possible when the exascale systems are delivered because all elements of the computing environment—applications, software, hardware, and systems—work seamlessly together.
ECP Project Office Offers Collaboration Tools and Training, Organizes First Annual Meeting
ECP R&D activities fall within one of four topical focus areas: Application Development, Software Technology, Hardware Technology, or Exascale Systems. A fifth component of the ECP—project management—supports the ECP in planning and executing these activities using best practices in project management for large, complex science projects. The ECP Project Office, located at Oak Ridge National Laboratory, helps deliver effective project management with integrity by providing communication products and methods, financial management, project controls, risk assessment, project reporting, quality assurance, procurement oversight, information technology and cyber security support, travel and meeting planning, project integration, and other services.
Recent activities include sourcing, delivering, and training for products from the Atlassian suite of software development, project management, and collaboration tools. More than 800 ECP team members are now using the Atlassian Confluence and JIRA tools to document R&D activities, milestones, and reports. The project office worked with ECP leaders earlier this year to organize and host the first ECP Annual Meeting in Knoxville, Tennessee, with more than 500 participants from the nearly 100 ECP R&D projects.
During a typical day in the project office, the staff may be tracking and executing contracts; updating descriptions of project milestones; developing reports for the Department of Energy; providing administrative and project support for focus area directors; responding to requests for help with project software tools; and planning future ECP events. Upcoming newsletters will include more details about some of these support activities.
Exascale Project Holds First Council Meeting with Industry Leaders
On March 6–7, the ECP Industry Council held its kickoff meeting with ECP senior leadership. The Industry Council’s 18 members are senior technology executives from prominent US companies, including manufacturers, service providers, and developers of simulation and modeling software used by industry.
The meeting was a launching point to introduce council members to the ECP and begin the process of collecting invaluable advice from members on developing the exascale computing ecosystem so that it can best help address industry’s key scientific and data challenges. ECP leadership presented project goals in application, software, hardware, and exascale systems development; and industry representatives asked questions and gave feedback about the project’s strategy for promoting US leadership in high-performance computing and the capabilities of the exascale technologies under development.
“At our first Industry Council meeting, the project team provided the private sector high-performance computing community an exciting and comprehensive overview of the goals and status of the ECP,” said ECP Industry Council chair Michael McQuade, Senior Vice President for Science and Technology at United Technologies Corporation. “In turn, council members offered insights on target applications, approaches to this kind of large-scale development project, and suggestions on overall industry communications. I was delighted by the open and productive dialogue.”
Co-executive directors of the ECP Industry Council are David Martin from Argonne National Laboratory and Suzy Tichenor from Oak Ridge National Laboratory. Members include senior executives from Altair Engineering Incorporated; ANSYS Incorporated; Cascade Technologies Incorporated; Chevron Energy Technology Company; Cummins Inc.; DreamWorks Animation; The Dow Chemical Company; Eli Lilly and Company; Exxon Mobil Corporation; FedEx Corporation; General Electric; General Motors Company; Mars Incorporated; Procter & Gamble Company; Tri Alpha Energy Incorporated; Westinghouse Electric Company; and Whirlpool Corporation.
Exascale Landing Pages Are Coming Online
Two of the ECP partner labs, Lawrence Berkeley National Laboratory and Los Alamos National Laboratory, have launched new exascale landing pages.
Berkeley Lab created a website highlighting the lab’s participation in the Exascale Computing Project. The site also describes other Berkeley Lab projects helping to prepare the DOE research community for computing in the exascale era.
LANL has a dedicated website to spotlight the lab’s work in many areas of exascale research, including its role as home of the ECP Co-Design Center for Particle-Based Methods: From Quantum to Classical, Molecular to Cosmological.
The other ECP core partner national laboratories—Argonne, Lawrence Livermore, Oak Ridge, and Sandia—are working to bring their exascale landing pages online.
Town hall introduces Argonne researchers to exascale
On January 24, members of Argonne’s ECP leadership team and staff collaborating in ECP efforts hosted a well-attended, lab-wide town hall meeting. Those in attendance learned more about potential collaborative opportunities within the ECP through presentations and an interactive poster session that followed the event.
ECP Meets Industry
View ECP Director Paul Messina’s March 7 presentation to the project’s Industry Council. The presentation is an overview of the history of high-performance computing at US Department of Energy national laboratories and an introduction to the project and its plan to design an exascale system by 2021.
Media Take on ECP Industry Council
EnterpriseTech covers the ECP Industry Council and the importance of exascale to US industrial and economic competitiveness.
ECP Industry Council Chair Addresses ECP Research Community
ECP Industry Council Chair, Michael McQuade of United Technologies delivers an inspiring speech to more than 450 ECP researchers and collaborators at the first ECP Annual Meeting.
Q&A: Science Applications for a New Era of Supercomputing
“Researchers will be able to more realistically simulate processes related to important energy applications,” says ECP Application Development director Doug Kothe of Oak Ridge National Laboratory. In this Q&A, Kothe and Application Development deputy director Bert Still of Lawrence Livermore National Laboratory discuss how science applications at exascale can improve our power grid, batteries, materials, and more. They also walk us through the ECP approach to application development.
Q&A: Designing Revolutionary Hardware Through Collaboration
“Application and system software developers do not usually have the opportunity to provide input into the specification of hardware technology requirements—but for ECP, they must,” says Jim Ang, Hardware Technology focus area director. In this Q&A, Ang discusses the new, mission-driven approach to hardware design for exascale.
First Request for Information Closes
The ECP recently closed submissions for its first Request for Information (RFI) “to provide the ECP and its associated laboratories with information for the creation of several requests for proposals related to the first US exascale system acquisitions.” The RFI, issued by the ECP in collaboration with the SC and NNSA lab supercomputing facilities was for informational purposes only and nonbinding, and covered characteristics of and technology advances that could be integrated into exascale systems between 2021 and 2025, as well as system providers’ descriptions of their software stacks and for software through 2025.