Plans for developing the NCAR Supercomputing Center (NSC) with Wyoming partners
NCAR is partnering with the University of
Wyoming and other entities within the State of Wyoming to develop a
supercomputing center dedicated to meeting the fast-growing, specialized
high-performance computing needs of Earth System scientists.
In partnership with the University of Wyoming (UW), the State of Wyoming,
Cheyenne LEADS, the Wyoming Business Council, and NSF, NCAR continues
making plans to develop the NCAR Supercomputing Center (NSC) in Cheyenne,
Wyoming. The vision for this collaborative project is summarized by:
- The primary purpose of the NSC should be to enable Earth System science discoveries.
- The facility should be dedicated to Earth System science problems.
- Because environmental problems know no boundaries, the NSC should serve to broaden participation in this geoscientific enterprise.
- The facility should be world class and built to last.
- The NSC should be energy efficient and as green to build and operate as practicable.
- Time is of the essence: many Earth System science questions have huge societal impacts and require dedicated computational resources to be made available as soon as possible.
This stated vision is well aligned with the NSF Strategic Plan and the NSF vision for cyberinfrastructure (CI). The project is motivated by the scientific needs of the Earth System sciences community and is being proposed in direct response to the exploding demand for both capability and capacity high-performance computing (HPC) resources from geosciences researchers. Whether because of a need for greater model resolution, increased model complexity, better statistics, more predictive power, longer simulation times, or a combination of these factors, Earth System investigators are clamoring for petascale computing, data analysis, and visualization resources and exascale data management capabilities.
The provisioning of such capabilities requires the availability of a large-scale computing center capable of handling the multi-megawatt heat loads of future systems. The basic size and infrastructure requirements for the NSC have been determined with these power demands in mind, and a preliminary conceptual design for the facility has been developed that will support the supercomputing needs of Earth System science researchers for the next two to three decades.
Plans for development of the NSC are fully aligned with NSF's larger CI vision and will directly contribute to the creation of a national petascale cyberinfrastructure. As proposed, the NSC will be a peer with other NSF Track-2 facilities and will serve as a "stepping stone" for Earth System science investigators to fully utilize NSF's multi-disciplinary, one-petaflop-sustained Track-1 facility.
Detailed FY2008 accomplishments
During FY2008, NCAR personnel continued their efforts to work with NSF and partners in Wyoming to develop a path forward for the NSC project. As part of an internal project review process, NSF commissioned a subcommittee of the NSF Geosciences Advisory Committee (GEO-AC) in fall 2007 to examine and make recommendations regarding the supercomputing needs of the atmospheric sciences community. This panel of community experts completed their work in winter 2008 and delivered their final report to the GEO-AC in April. Among the panel's findings were:
- "The National Center for Atmospheric Research (NCAR) plays a critical role in providing high-end computing facilities, leadership in developing community models, and data curation for the ATM research community, and will continue to do so in the future."
- "The computing needs of the ATM research community are rapidly outstripping the ability to provide them within the NCAR Mesa Lab."
- "Any additional computing facilities aimed at supporting ATM science must be tightly integrated with the existing facilities provided by NCAR, e.g. by means of high-speed networks and shared file systems."
Key recommendations made by the panel included that NSF should "continue to support the work that NCAR is doing for the community in providing computing facilities ... community model development, and data curation" and "provide a new Track-2-class facility, dedicated to ATM researchers, through an open competition with a highly specific set of ATM requirements."
In response to the subcommittee findings and recommendations, NSF made the decision in early summer 2008 to support plans developed by NCAR, UCAR, and Wyoming for construction of the NSC. In late August, NSF, UCAR, and NCAR personnel agreed on a multi-step process to pursue project initiation and approval. NCAR, UCAR, and the UW are proceeding with the release of a request for information (RFI) for architecture and engineering (A&E) services. The RFI release will precede the preparation and release of a formal request for proposals (RFP) as part of the A&E firm selection process. In parallel with this selection process, NSF will utilize external and internal guidance to define a formal project process. In late FY2009 following completion of the initial facility design effort, NCAR will prepare and submit a project proposal to NSF. This proposal will be reviewed as part of the multi-step project review and approval process to be put in place by NSF.
Impacts
As envisioned, the NSC will play a key role in providing HPC resources to the Earth System sciences community for many years to come. Planning and development of this large, complex project has required six years of effort on the part of NCAR personnel, and roughly four more years of work will be required before NSC supercomputing resources will be available for use by the scientific community. NCAR intends to fully integrate the NSC into the NSF TeraGrid follow-on, and the NSC development effort will be conducted within the guidelines of the NSF Strategic Plan and the NSF cyberinfrastructure vision. Earth System science researchers are faced with immense challenges to effectively utilize petascale systems, and the availability of large-scale parallel systems via the NSC -- in combination with access to software engineering and parallel algorithm research expertise -- is needed to ensure that investigators can prepare their codes and simulations to run effectively on NSF Track-1 computing resources.
This work is supported by NSF Core funding.