Call for Leadership-Class Computing Facility Application Partners

Synopsis

The call describes an opportunity for science teams with current or future requirements exceeding those routinely available on today’s large-scale scientific computers (e.g., supercomputers) to participate in establishing design requirements for a new leadership-class computing facility being planned by the National Science Foundation and the Texas Advanced Computing Center. Partner science teams will provide application codes (or workflows if multiple codes are needed) and “grand challenge”-class problems that will drive the design of the NSF LCCF.

Background

Planning is underway for NSF’s Leadership Class Computing Facility (LCCF), a large-scale computing and data resource that is anticipated to be funded through the NSF’s Major Research and Equipment Facility Construction process. When commissioned in 2025, the LCCF will comprise an ecosystem for very large-scale computing in support of the Foundation’s mission to promote our nation’s progress in science. Taken together – with facilities, staff expertise, partner institutions, storage, data analysis, advanced networking, and one or more supercomputers – the LCCF will deliver a ten-fold or more time-to-solution performance improvement for science users over the performance of NSF’s Frontera supercomputer.1 To support this, LCCF will provision unique computational services, distributed sets of resources, and expertise to transform our nation’s S&E research with novel CI and discovery workflows. The LCCF will operate for at least ten years to provide long-term sustained support to the research community.

The design of the facility is informed by the requirements of the community of science teams that will make use of it. The LCCF has engaged in a community-driven process to develop requirements, conducting workshops on prospective science challenges (such as the one described here) that have led to an emerging plan for the facility. Many of the future requirements will be expressed through a set of “Characteristic Science Applications” – application and “grand challenge”-class science problems – identified by the community of current and emerging large scale scientific computing users. The CSAs could include but are not limited to applications that required extreme scales, workflows that integrate data intensive and possibly distributed analytics, and applications that explore innovative uses of AI/ML. The CSAs are the key to a successful design that effectively serves its science users, and therefore we anticipate developing the set of CSAs over time through a close partnership with the science community. The purpose of this call is to establish the initial partnerships that will begin the CSA selection process. Partners will receive funding over multiple years to refine their chosen applications to run on the future LCCF architecture.

Description of the Opportunity

Science teams who today, or who plausibly may in the near future, require large-scale scientific computing to advance the state of the art in their disciplines are invited to submit an application form describing a grand challenge problem to be solved in their discipline, why LCCF resources will be needed, and what methods/codes will be used to solve the problem.

Because a substantial effort will be required of each team, the LCCF will provide direct funding to partner teams as well as a dedicated staff expert to assist in analyzing the existing code (or set of codes), completing a gap analysis and code changes, and running the challenge problem on the new platform. LCCF anticipates funding:

  • Up to 20 teams will be selected from the initial applications to undergo a technical evaluation with the LCCF team to construct a final proposal.
  • 10-15 teams at $120-$150k per year for the first year of study and design, with a commitment from the science team to collaborate with the LCCF project to improve the code (or set of codes) and prepare it for the candidate architecture. Teams making sufficient progress may be renewed for a second year of funding at the same level during final design.
  • 6-10 teams will enter the construction phase of the LCCF project and be funded for approximately 30 months as the CSA is demonstrated on the LCCF’s HPC resource(s).

Partner Selection Criteria

At each stage of the process (initial selection through final capability demonstration) the LCCF team will evaluate the potential of partner applications along three dimensions:

  1. Significance: Is the problem scientifically significant?
  2. Representation: Does the inclusion of this problem add to our representation of scientific codes that the LCCF facility will likely need to serve?
  3. Suitability: Given the current state of our ability to solve this problem, does it make sense to include this problem in the suite?

Process Description and Timeline

  • Between February and April 2021, up to 20 teams will be selected for an in-depth technical assessment.
  • The result of the technical assessment and additional materials from the science teams will be used to evaluate each application to determine which projects will receive funding to continue developing their codes in conjunction with LCCF.
  • In June 2021, the NSF will review the results of this process.
  • Following NSF concurrence, 10-15 selected science teams will be funded to collaborate with the LCCF project to improve the code and prepare it for the candidate architecture.
  • In FY24, 6-8 of the teams that have made sufficient progress will continue to be funded (at levels previously established and with the same incentives of support and allocation) as part of the LCCF Construction phase until LCCF production operations start (approximately 30 months). During this period the code will continue to be jointly developed and the target problem will be run during the early science period on the LCCF early-access system.

Each team that participates in the process will exit with an improved science application and new computational insights into their code; teams selected for the entire process will have substantially improved applications that achieved new science results on one of the newest and fastest supercomputers in the world.

How do I apply?

  • Consideration of applications will begin January 31st, 2021. Additional applications will be accepted up through February 26th, 2021.
  • Apply online at lccf.tacc.utexas.edu/application.

The submission provides an opportunity for potential partners to describe their application code, the significance of the challenge problem to be addressed, why leadership-class resources will be needed, and what methods/codes will be used to solve the problem.

Questions? Email lccf-community@tacc.utexas.edu.

1Frontera is the successor to Blue Waters and NSF’s current petascale system.

 

APPLICATION CLOSED