NSF Advanced Computing Systems & Services: Adapting to the Rapid Evolution of Science and Engineering Research

The following is a limited submission opportunity with only ONE proposal as lead institution is allowed. If you are interested in this program please email Michelle Wonsey a single PDF by July 31, 2019 with the following pre-proposal requirements:

1.      A two-page summary that must include your research/project:

a.      goals
b.      objectives
c.       methods, and 
d.      a short statement of competitiveness (i.e., what you think will discriminate your proposal from the competition, including anything you have done to pre-position yourself/your team for this funding opportunity)

2.      A pro forma budget, including any cost-sharing and facilities requirements and how you plan to meet them; use template: https://www.umass.edu/research/form/pro-forma-budget-template

3.      A short-form CV for the PI and each senior staff person

4.      Current and Pending Support of PI’s

Program Description:

The intent of this solicitation is to request proposals from organizations willing to serve as service providers (SPs) within the NSF Innovative HPC program to provide advanced CI capabilities and/or services in production operation to support the full range of computational- and data-intensive research across all of S&E.

To increase the Nation’s capacity for transformative S&E discoveries, NSF is interested in continuing to diversify and evolve its portfolio to take advantage of new technologies and services that include capabilities addressing emerging computational- and data-intensive S&E research topics, workflows, and communities, while expanding opportunities for participation by a broader range of potential SPs.

This competition emphasizes the provisioning of an ecosystem of advanced computational resources and services that is responsive to the dramatic increase in the number and nature of applications using NSF-funded resources. Proposals are requested for advanced CI that will deploy capabilities and services, including composable services, to address the increase in demand for computation and data analytics resources in the S&E research community, as well as explore novel paradigms for enabling transformative S&E discoveries.

This solicitation explicitly focuses on the growing scale and diversity of the S&E community, the changing nature of S&E research requirements, as well as the rapidly evolving CI landscape, with the overarching goal of supporting transformational S&E discoveries.

An important aspect of the current solicitation is that funded projects must provide CI capabilities and/or services that demonstrate high degrees of stability and usability during the period of production operations available to the S&E community. NSF strongly urges the community to consider expanding the range of possibilities in enabling S&E communities to leverage the power of computation for transformative research, and to think broadly about the nature and composition of the CI ecosystem. Such consideration may include, but is not limited to, ease of access to proposed systems/services by new S&E communities; new capabilities that will open up new methods and paradigms for S&E discoveries; federated approaches with opportunities for leveraging the increasing availability and capabilities at the network edge (including campuses); and composable services provisioning virtualized on-premise computing infrastructure and commercial cloud services.

The current solicitation is intended to complement previous NSF investments in advanced computational infrastructure through provisioning resources in two categories as described below.

Category I – Capacity Systems

Resources proposed in this category are intended to be operational deployments of production computational resources that will provide maximum capacity and throughput to support the broad range of computation and data analytics needs in S&E research. The proposed resource must be clearly motivated by the current and future demand for computational and data analytics capacity in the broad and diverse S&E research community. This category particularly targets capabilities and/or services for small- to mid-scale jobs (from one to a few thousand cores per job) across broad areas of S&E, including support for “long-tail science” applications, as well as new classes of applications, such as artificial intelligence/machine learning/deep learning applications. Proposers are encouraged to explore novel models that leverage federated and/or distributed resources, regional and/or campus supported resources, and/or commercial cloud services.

Competitive proposals in Category I must address the following requirements in the Project Description:

  • A clear plan for provisioning a resource or service that addresses the current and future demand for computation and data analytics capacity in the broad S&E computational research community.
  • A forward-looking plan for engagement with other NSF supported efforts in advanced CI where possible, such as domain-specific centers housing software, sensors or instrument data that will enable new advances in S&E research.
  • A persuasive articulation as to how the resource will support less traditional computational S&E communities if appropriate and explore models of engagement with campus-supported CI.
  • A description of how the resource will support S&E research communities that require a national-scale, on-demand, compute and data-analytics resource with a flexible and user-friendly software environment.
  • A clear plan of operations for the project duration with a clear set of operation and science impact metrics to ensure the resource will be an asset for the nation’s S&E research community.
  • A detailed risk-mitigated deployment plan to ensure that the proposed resource will be in production operations and available for allocation to the open S&E research community no later than 12 months from the time of award.
  • A comprehensive set of system-level performance and reliability metrics that will be used by NSF for acceptance of the resource or service.

Relevant parameters contributing to the comprehensive technical description of the proposed system may vary with the nature of the resource. However, all description of the proposed resource must closely adhere to the guidelines provided in section V.A. Proposal Preparation Instructions.

Category II – Innovative Prototypes/Testbeds

Resources proposed in this category will be initially deployed as a prototype/testbed system supporting S&E research through delivery of novel forward-looking capabilities and services. Resources proposed in this category can represent the deployment of new technologies, system architectures, or usage modalities at scale, with plans for developing a national S&E user community that will benefit from the proposed capabilities. Proposed systems could include novel processor architectures supporting artificial intelligence applications, distributed systems leveraging edge devices, domain-specific architectures and technologies, such as but not exclusively, reconfigurable and/or software defined systems, systems designed for streaming data and/or real-time processing, etc.

Proposers must clearly define the target classes of S&E applications that will be enabled, as well as a clear plan for ensuring the widespread adoption by these classes of applications on the proposed capabilities and/or services. While the resources in this category may initially include prototypes/experimental testbeds, proposers are expected to present a clear near-term plan for transitioning to high-availability production services broadly available and allocable to the S&E community through open peer-reviewed processes during the final 24 months of the project award period. It is also expected that the initially-deployed prototype/testbed will include active engagements with S&E researchers, and these engagements will be reviewed by NSF in its evaluation of the system. Clear science impact metrics for measuring the performance of the proposed system are required.

Competitive proposals in Category II must address the following requirements in the Project Description:

  • A clear plan for provisioning innovative computational and data analysis capabilities or services that will enable new methods and paradigms in support of transformational S&E discoveries;
  • A compelling description of how the proposed capabilities or services will address future demand for computation and data analytics capabilities [see the XD Metrics Service (XMS) Workload study] in S&E research;
  • A persuasive set of S&E use cases, including quantitative analysis through benchmarks, that clearly motivate how the resource will expand the range of S&E applications that can be currently tackled using existing NSF-funded Innovative HPC resources;
  • A forward-looking plan for engagement where appropriate with other NSF-supported efforts in advanced CI, such as with domain-specific centers housing software, sensors, or instrument data, to enable new advances in S&E research;
  • A clearly defined set of target S&E application classes that will be enabled, as well as a clear plan for ensuring the widespread adoption by these classes of applications on the proposed capabilities and/or services;
  • A clear plan of operations for the project duration, with a detailed set of engagement activities with the S&E research community, to optimize the use of the resource, facilitate application and user transition during the initially-deployed prototype/testbed system phase, and ensure that the resource evolves to a high-availability production utility for a national community of S&E users;
  • A detailed risk-mitigated deployment plan to ensure that the proposed resource will evolve to high-availability production services broadly available for allocation to the open S&E research community in the final 24 months of the award period; and
  • A comprehensive set of system-level performance and reliability metrics that will be used by NSF for acceptance of the resource or service.

Project duration for both categories will depend on the nature of the resource to be deployed. Resources may be deployed in one step near the beginning of the award period, or progressively in a series of steps or phases during the award period.