NSF Cyberinfrastructure Resource Allocations Policies
(Updated: October 2005)
2. The Basics of Allocation Requests and Awards
2.2. Appropriate Uses for Allocations
2.3.1. Categorizing Multi-Year Requests
2.4. Multiple Proposals and Awards
2.4.2. Multiple Proposals vs. Collaborative Proposal
3. Submission Types and Proposal Format
4. Proposal Preparation Guidelines
4.2. Calculating Resource Requests
4.2.1. Definition of the Service Unit
4.2.2. Recommended Use Guidelines
4.2.3. TeraGrid Resource Requests
4.4. Recommended and Optional Attachments
4.5. Samples of Successful Proposals
4.6. Allocation of Non-CPU-oriented Resources
5. Proposal Processing and Review
7. Further Details on Eligibility
7.1. Special Eligibility Rules
7.1.3. Non-Profit, Non-Academic Organizations
7.1.4. For-Profit Organizations
7.1.5. State and Local Governments
7.1.6. Unaffiliated Individuals
7.2. Eligibility of Other Federal Agencies and FFRDCs
7.2.2. Federally Funded Research and Development Centers (FFRDCs)
This document provides policy guidance and instructions for preparing proposals for resource allocations from National Science Foundation Cyberinfrastructure resources. Currently, such resources are at the National Center for Supercomputing Applications (NCSA), the San Diego Supercomputer Center (SDSC), and the TeraGrid Resource Partners (formally, the NSF Extensible Terascale Facility).
The best way to obtain access to NSF resources depends on your eligibility and resource requirements. Startup (development) allocations are accepted and reviewed continually. The Medium Resource Allocation Committee (MRAC) reviews medium-sized requests quarterly, and the Large Resource Allocations Committee (LRAC) considers the largest requests twice each year. This panel of computational scientists reviews proposals and recommend awards for all NSF resources.
This update contains no major policy revisions. The document has been updated to reflect current nomenclature and clarify existing policies.
This guide is intended to provide the information necessary to submit proposals for allocations of time on NSF resources. Greater details on the process are provided later in this document.
In general, to apply for an NSF resource allocation you must be a researcher or educator at a U.S. academic or non-profit research institution. A principal investigator (PI) may not be a high school, undergraduate, or graduate student; a qualified advisor must serve in this capacity. A post-doctoral researcher is eligible to serve as PI.
Researchers need not be supported by NSF grants; PIs may have support from any funding agency or funding source.
A scientist, engineer or educator who has a joint appointment with a university and a federal agency may submit proposals using this university affiliation. In most cases, scientists, engineers or educators solely employed by federal agencies or Federally Funded Research and Development Centers (FFRDCs) not supported by NSF are ineligible for awards.
Generally, the eligibility guidelines outlined in the current NSF Grant Proposal Guide apply. See §7 (Eligibility) for further details.
Individual Investigators: Most allocations support single PIs and possibly small research teams, all of whom work closely together as evidenced, for example, by co-authored papers.
Large Research Collaborations: Allocations of this type are characterized by a single PI representing a large group of collaborating researchers with co-PIs who represent primary projects within the overall collaboration. The proposal is jointly developed and the division of awarded resources is by consensus amongst the principals on the proposal.
Community Projects: Allocations of this type are intended to support resources usage by large-scale, funded, community projects. In these cases, there is frequently already a mechanism for allocating community or project resources (e.g., an instrument such as a telescope or detector), which will also be used to allocate to individual investigators’ time granted to the community project.
The internal processes for review and internal allocation of any awarded resources must be clearly documented in the proposal. The Allocations Officers and/or an NSF representative will review these processes to determine acceptability in satisfying NSF merit-review criteria. In addition, information should be provided on the grants supporting the community activity, if any.
Community Services: Proposals of this type are intended to support projects that provide services to a large community of users who are typically not collaborating with the PI of the submitted proposal. An example of such a project would be an application portal service providing access to software and cycles to a community of users via the developed service.
This type of proposal needs to describe the services provided, the methods used and the expected consumption of resources. It is anticipated that most such services will consume resources under a single or very limited number of logins, but that the service itself will provide some tracking of usage by individuals making use of the service, and this should be reported in renewal requests for resources, progress reports, and end-of-project reports.
The deadlines and review process for a proposal depends on the amount of resources requested in terms of CPU-hours, or service units (SUs). (See §4.2.1 for a more detailed definition of SUs.)
Large Resource Allocations Committee (LRAC): The LRAC reviews proposals requesting more than 200,000 SUs per year across all NSF resources.
Medium Resource Allocations Committee (MRAC): The MRAC reviews proposals requesting between 10,000 and 200,000 SUs per year across all NSF resources.
Development Allocations Committees (DAC): Development allocations are requests for up to 10,000 SUs on an individual resource (30,000 SUs on TeraGrid). These allocations target new users planning to submit more substantial proposals in the future, users who have very modest computational requirements, or faculty intending to use resources for classroom instruction or projects.
It is generally expected that multi-year proposals (see §2.7, Award Duration) will have annual resource requirements that fall within the same request level. If this is not the case, PIs are encouraged to submit single-year awards.
Activities encompassed by multi-year proposals should be mature enough to begin research runs at the start of the award period. Significant start-up efforts (code development and testing) should be completed via DAC awards or single-year awards prior to submission of a multi-year proposal.
In general, an individual PI should submit only one LRAC, MRAC or DAC proposal and can have only one active LRAC, MRAC, or DAC award. Several computational experiments can be combined in a single proposal, however. The resource request for each experiment must be justified, and the LRAC, MRAC and DAC award-size limits apply to the aggregate request.
The single-award and single-proposal rule ensures that the award-size limits are not circumvented. PIs are expected to submit new or renewal MRAC or LRAC proposals to continue their research after the expiration of their current awards.
There are several exceptions to this rule:
To add SUs to or extend the end date of a current award, see §6.3, Supplements, or §6.4, Extensions, for information.
As a general rule, closely collaborating researchers are encouraged to submit a single collaborative proposal rather than several individual-scale proposals. For example, a PI and associated post-doctoral researchers; investigators supported by the same funding award; and researchers in the same lab group should submit a proposal describing and justifying the various activities being undertaken. For guidance on whether multiple proposals or a single, collaborative proposal is appropriate in a specific case, please contact any of the Allocations Officers (§6.5).
The deadlines for submitting a proposal depend on the number of SUs requested.
Proposals are due approximately six weeks prior to the meeting at which they will be reviewed. The LRAC meets twice a year, in March and September, and makes awards effective April 1 and October 1, respectively. The MRAC meets quarterly and makes awards effective January 1, April 1, July 1, and October 1.
A detailed schedule of proposal deadlines for upcoming MRAC and LRAC meetings is posted on the CI-Partnership.org web site at http://www.ci-partnership.org/Allocations/.
Development Allocations requests can be submitted at any time to the appropriate Development Allocations Committee (DAC).
All LRAC, MRAC and DAC allocations requests must be submitted through the Partnerships Online Proposal System (POPS) at https://pops-submit.ci-partnership.org/. Hardcopy proposals are not accepted. POPS accepts PDF, Microsoft Word, and PostScript files; PDF files are the preferred format. Complete information on the POPS system is available at http://www.ci-partnership.org/Allocations/.
Single-Year Awards: Most requests and allocation awards of NSF resources are made for a 12-month period. PIs can submit renewal requests for allocations in subsequent years. All DAC awards are for one year.
Multi-Year Awards: Experienced PIs can submit requests for awards of NSF resources for periods greater than 12 months and typically for several years. These proposals directly support a scientific grant with multi-year funding; allocations may be requested for the duration of that support. The submission must indicate the supporting grants for the computational activities. POPS allows PIs to specify that the request is multi-year. Multi-year requests are held to the same review criteria as single-year awards.
While a commitment is made to provide resources for the life of the award, actual allocations are made for 12-month periods. Continued allocations are not automatic. Each year, progress reports and streamlined subsequent year requests are required. Progress reports should indicate usage, accomplishments, publications and similar progress by the research team. Continued allocation will depend on the MRAC or LRAC determination of sufficient progress.
IMPORTANT: It is highly recommended that all new PIs submit single-year requests and renewals until they are familiar with the proposal format, have received favorable reviewer feedback, and can demonstrated solid progress with a single-year award or awards.
At the end of a 12-month award period, single- and multi-year projects forfeit any unused SUs. However, projects that encounter problems in consuming their allocations, such as unexpected staffing changes, can request an extension to their award date. See §6.4, Extensions.
The general format of the proposals is based on the NSF Grant Proposal Guide (GPG) and therefore the guidelines should be familiar to PIs submitting requests. All proposals should conform to the instructions provided in the GPG and in this document. Proposals that are not consistent with these instructions may be returned without review.
DAC requests do not require a formal proposal document. PIs should fill out the POPS forms as completely as possible. An abstract of the work to be conducted and a CV for the PI are required.
The allocations process and the POPS system recognize five types of submissions. Each type of submission has its own proposal document page limit and requirements.
The following elements are entered directly into POPS web forms as part of the submission. Since they are entered in the POPS forms, these elements need not be repeated within the proposal document.
The main body of the proposal discusses the work, justifies the resource request and provides any additional information. The main body of the proposal must adhere to page length rules. See §4, Proposal Preparation Guidelines, for help in creating your proposal.
The following page limits apply to the main document of the proposal. If page limits are exceeded, the allocation committees reserve the right to return the proposal without review. Reviewers do not count figures (within reason) and references/bibliographies against the page length.
Attachments (such as CVs) do not count against the page limit. However, PIs are strongly discouraged from using attachments to circumvent the page limits.
| Proposal Document | Page Limit |
| LRAC Multi-Year Proposal | 15 pages |
| LRAC Single-Year Proposal OR MRAC Multi-Year Proposal |
12 pages |
| MRAC Single-Year Proposal | 10 pages |
| LRAC or MRAC Renewal Proposal OR LRAC or MRAC Justification Request or Supplemental Request OR Annual progress report and request for subsequent year resources in multi-year proposals |
7 pages |
While readability is of greatest importance, the following minimum requirements exist.
Margins: Proposals must have 2.5 cm (1 inch) margins at the top and bottom and on each side.
Fonts and Spacing: The type size used throughout the proposal must conform to the following three requirements:
Pagination: Each section of the proposal uploaded as a separate file must be individually paginated before uploading to POPS. Pagination is not provided by the POPS system.
File Format: POPS accepts PDF, Microsoft Word, HTML, and PostScript file formats. PDF files are recommended.
The allocation award process is very competitive. The review of proposals generally follows the review criteria stated in the NSF Grant Proposal Guide. However, the single most important review criterion is the justification of the resource request.
Inadequate justification for requested resources is often the reason for reduced or denied awards.
The bulk of a proposal for NSF resources should focus on the resource request, the appropriateness of the computational methodology, the algorithmic capabilities that make this problem a good fit for the requested machines and the reasons for the number of runs being proposed.
A proposal will clearly and succinctly state the importance of the research to be conducted. Research projects typically have separate financial support through an independently reviewed proposal, and therefore, the scientific merit of the research has had independent review and has been deemed worthy of support. Thus, it is important to enter Supporting Grant information in the POPS forms.
In the cases where other support for the proposed work is not noted, it will be assumed that such support does not exist, and the allocations committee will conduct some level of review of the science.
Resource requests should specify the codes used, the number and type of runs to be made and the resources required for these runs (SUs, number of processors and/or other relevant hardware characteristics). These should be clearly linked to the scientific goals and the proposed computing methods.
Requests for NSF resources need to be detailed and justified in the proposal and also entered in the POPS forms.
CPU resources are requested in the form of Service Units (SUs). In general, an SU is equivalent to either one CPU-hour, or one wall-clock-hour on one CPU, of the system of interest. The exact definition of an SU on each platform is defined in the technical documentation associated with each resource available through the listing of NSF resources at http://www.ci-partnership.org/Resources/ciresources.html.
For each of the resources available for allocation via the NSF allocations process, there is a statement of recommended use. In addition, requests for TeraGrid Cluster time and TeraGrid Wide Roaming Access have specific usage implications.
Links to recommended use guidelines for NSF resources are available from http://www.ci-partnership.org/Resources/ciresources.html. PIs are advised to review these guidelines before submitting requests for specific resources.
PIs who need the special capabilities of particular TeraGrid resources can request those specific resources in their proposals. For specific resource requests, the guidelines in §4.3, Review Criteria, apply.
In addition, TeraGrid resources can also be allocated as a collective resource. An allocation for “TeraGrid Wide Roaming Access” provides access to all TeraGrid resources.
For TeraGrid Wide Roaming Access, in addition to the standard resource justification, proposals should also describe how the research would take advantage of roaming access capabilities. Estimates of SUs a PI expects to use on particular systems are useful for the sites’ planning purposes, but not required. However, see the TeraGrid recommended use guidelines for the implications of these requests.
A successful proposal will clearly describe three criteria to permit evaluation of the resource justification: Computational Methodology, Appropriate Use of Resources, and Efficient Use of Resources. In addition, PIs should describe their local computing resources and other supercomputer resources available to them.
Computational Methodology: The choice of applications, methods, algorithms and techniques to be employed to accomplish the stated objectives should be reasonably justified. While the accomplishment of the stated objectives in support of the science is important, it is incumbent on proposers to consider the methods available to them and to use that which is best suited.
Appropriate Use of Resources: The resources chosen must be an appropriate match for the applications and methodologies to be used and must be in accordance with the recommended use guidelines for those resources.
Efficient Use of Resources: The resources selected must be used as efficiently as is reasonably possible. To meet this criterion, performance and parallel scaling data should be provided for all applications to be used along with a discussion of optimization and/or parallelization work to be done to improve the applications.
Local Computing Environment: Proposals should describe briefly the local (e.g., campus or lab) computing environment available to a research team and, most significantly, how NSF resources will provide capabilities beyond those of local resources.
Other Supercomputer Resources: If a PI has access to other supercomputer resources or may appear to have such access, as with researchers who are affiliated with or are collaborating with other agency-supported facilities, the proposal should describe why those resources are not available for the proposed project.
Prior Progress: For renewal requests and continuations of multi-year awards, reviewers also consider a PI’s success at using prior allocations and achieving meaningful results. This information should be detailed in the progress report (see §4.4).
Attachments are submitted along with the main proposal document. No attachments are required, and some PIs choose to include all or some the following information as part of a single file. When doing so, PIs should take care to clearly indicate where the main proposal body ends and the additional information begins.
As a general rule, attachments should augment information presented in the main proposal document, but should not be required reading in order for reviewers to make their recommendations.
To assist PIs in writing their proposals, samples of well-written and successful proposals have been made available as models for others to follow.
MRAC and LRAC members have expressed the view that these were excellent examples of how an allocation request should be presented. In particular, the reviewers noted that in each case:
Currently, resources other than CPU time are not allocated. However, it is anticipated that the allocation of non-CPU resources (such as data storage or dedicated networking resources) may be required in the future and this document will be updated to reflect any such changes at that time.
PIs who have a need for non-standard data resources, such as database or data-collection hosting, can request access to specialized data resources. See http://www.ci-partnership.org/Resources/dataresources.html.
The LRAC meets twice per year and the MRAC meets four times per year to review proposals. The committees consist of volunteers who are selected from the faculty and staff of U.S. universities, laboratories and other research institutions. All of the committee members have expertise in some area of computational science or engineering and serve a term of 2-5 years.
Each MRAC proposal is assigned to 2-3 committee members for review; LRAC proposals are reviewed by 3-5 members. In addition to providing their own reviews, committee members can also solicit a supplemental external review of the proposal. After having had several weeks to review the proposals, the entire committee convenes to discuss the relative merits of each proposal and award time based on the availability of resources.
The effective dates for awards are based on when they are reviewed. For proposals subject to MRAC or LRAC review, awards are for a one-year period and are effective on January 1, April 1, July 1, or October 1, whichever is the nearest date following the review meeting. DAC awards are for a period of one year beginning at the soonest practical date following review of the request.
PIs will be notified of the results of the review of their proposal via a formal notification letter within three weeks following the review meeting.
All reviews are considered confidential and are only made available to the following:
Every effort is made to avoid conflicts of interest. Committee members are not allowed to review or be present for the discussion of proposals from their home institution; former students, postdocs, or advisors; or current and recent collaborators.
If in the opinion of a PI, a certain individual has a conflict of interest, the PI may request that the individual not act as reviewer on their proposal or potential subsequent appeal. The allocations coordinator will consider such requests for the particular review. Such a request should be sent to allocations@ci-partnership.org for both MRAC and LRAC reviews.
PIs who wish to appeal the decision of a review board (LRAC or MRAC) may do so by submitting a “justification.” The justification should be submitted via POPS within four weeks of the PI being notified of the committee results.
The PI may use the justification to supply additional information or clarification requested by the reviewers in order to make an award. A PI may also use the justification to specify why, in the PI’s opinion, the reviews of the original proposal and the resulting allocation decision were incorrect or unfair.
If possible, the allocations committee members who reviewed the original proposal will consider the justification. If there is no response after five weeks from the reviewers, the associated Allocations Officer will make a determination as to any award to be made in addition to any initial award, if resources are available. If the resources to be awarded are not available on the requested resource, the allocation may be made on an alternative resource. Alternatively, the allocation could be made at the beginning of the next allocation cycle on the requested resource.
If a satisfactory outcome has not been reached after the initial appeal, other board members and/or new external reviewers will, at the discretion of the Allocations Officer, be asked to review the original proposal. This third round of reviewing is final.
PIs are allowed to request a pre-award advance of up to 10% of the resources requested in the proposal in anticipation of favorable review. Any usage of this pre-award allocation will be debited against the actual allocation made.
PIs can request that awarded time be transferred from one platform to another, subject to availability on the target resource. Such transfers are subject to conversion of SUs according to established weighting factors. PIs can submit such requests to a site’s allocation officer (for transfers between systems at one site) or to allocations@ci-partnership.org (for transfers between sites).
For projects proceeding more rapidly than anticipated or requiring more resources than anticipated, PIs may submit a Supplemental Request for additional SUs. The Supplemental Request is an add-on to the original proposal request and, if awarded, is valid only for the period of the original award. Supplemental Requests must be submitted via the POPS system.
For projects proceeding more slowly than anticipated, due to staff changes or other factors, PIs may request an extension of up to six months to the expiration date of an allocation. This can be accomplished by submitting a request via the POPS system. After logging in, select the "Extension" proposal type. The request will indicate the project PI, the length of the extension and aF brief description of the reason for the extension.
An acknowledgement of support from the appropriate center or the TeraGrid should appear in a publication of any material, whether copyrighted or not, based on or developed with NSF-supported advanced computing resources. For suggested language, see http://www.ci-partnership.org/Allocations/acknowledgment.html.
In order to facilitate the allocations process, four Allocations Officers have been named to represent NCSA, SDSC, PSC and the other TeraGrid resource providers. The Allocations Officer for each site is responsible for assuring that the NSF allocations policies and procedures are followed and for making decisions in cases of policy questions. This set of persons is also responsible for keeping this document current as the program evolves.
Currently, the Allocations Officers (along with back-up persons) are:
| NCSA | John Towns |
Mike Pflugmacher |
| SDSC | David Hart |
Nancy Wilkins-Diehr |
| PSC | Ralph Roskies |
Ken Hackworth |
| TeraGrid | Ray Bair |
In general, to apply for an NSF cyberinfrastructure allocation you must be a researcher or educator at a U.S. academic or non-profit research institution. A principal investigator (PI) may not be a high school, undergraduate, or graduate student; a qualified advisor must serve in this capacity. A postdoc is eligible to serve as PI. If your institution is not a university or a 2- or 4-year college, special rules may apply.
Generally, the guidelines outlined in the current NSF Grant Proposal Guide, and presented here, apply.
NSF does not normally support research or education activities by scientists, engineers or educators employed by federal agencies or Federally Funded Research and Development Centers (FFRDCs). However, a scientist, engineer or educator who has a joint appointment with a university and a federal agency (such as a Veterans Administration Hospital or an FFRDC) may submit proposals through the university and may receive an allocation if he/she is a bona fide faculty member of the university, although part of his/her salary may be provided by the federal agency.
Under unusual circumstances (see §7.2 below), other federal agencies and FFRDCs may submit requests. Preliminary inquiry should be made to the appropriate center before preparing such a proposal for submission.
The requirement remains, as stated above, that the PI be a researcher or educator at a U.S. institution. Allocation requests may be made for projects with a domestic PI and a foreign component. For purposes of this policy, a "foreign component" is defined as performance of any part of the project outside the U.S. either by the PI or by a researcher or researchers employed by a foreign institution. The NSF will not make an award to a PI that does not have a substantive role in the project and would simply serve as a proxy for a foreign researcher.
Collaborative proposals involving non-U.S. researchers are encouraged as long as they include substantive intellectual collaboration with U.S. researchers. In general, foreign collaborators are eligible to participate in projects awarded an allocation and to make use of that allocation in a manner consistent with the request.
Allocations requests involving foreign collaborators will be evaluated using the standard review criteria. In addition, as part of the review process and award decision, the allocation request will also be assessed with respect to whether it presents special opportunities for furthering research programs through the use of unusual talent or resources in other countries that are not readily available in the U.S. or that augment existing U.S. resources.
Independent museums, observatories, research laboratories, professional societies and similar organizations in the United States that are directly associated with educational or research activities are eligible.
U.S. commercial organizations, especially small businesses with strong capabilities in scientific or engineering research or education may apply for an allocation. An allocation request from a commercial organization may be granted when the project is of special concern from a national point of view, special resources are available for the work, or the proposed project is especially meritorious. The NSF is interested in supporting projects that couple industrial research resources and perspectives with those of universities; therefore, it especially welcomes requests for cooperative projects involving both universities and the private commercial sector.
State educational offices or organizations and local school districts may submit allocation requests intended to broaden the impact, accelerate the pace, and increase the effectiveness of improvements in science, mathematics, and engineering education in both K-12 and post-secondary levels.
Scientists, engineers or educators in the U.S. and U.S. citizens may be eligible for support, provided that the individual is not employed by or affiliated with an organization and:
Unaffiliated individuals should contact the appropriate center before preparing a proposal for submission.
The Foundation has express statutory authority to support research performed by other Federal agencies. Nonetheless, NSF support of research at other agencies could have the effect of augmenting their budgets beyond what the President and the Congress have considered justified and could alter the allocation of national resources previously approved. Moreover, the Foundation believes that plurality of support for research has contributed greatly to the extraordinary success of U.S. science. The Foundation therefore tries to avoid actions that could lead other agencies to diminish support of research relevant to their missions, particularly basic research. ("Federal agencies" as used here refers to all agencies of the Federal Government, including the Smithsonian Institution. However, awards administered by the Smithsonian for Smithsonian research associates who are not and have not been regular Smithsonian employees are treated as awards to a nonacademic non-profit organization.)
The considerations above apply to Federally Funded Research and Development Centers (FFRDCs). Although FFRDCs are not federal agencies, each of them was essentially created by a federal agency and receives the preponderance of its resources (70 percent or more) from that particular agency. NSF support of research at a non-NSF FFRDC could effectively augment the other agency's budget and could lead to diminished support of research, particularly basic research, relevant to NSF’s mission.
A list of FFRDCs in FY2004 is available at http://www.nsf.gov/sbe/srs/nsf04309/.
NSF supports the following FFRDCs:
The NSF ordinarily does not encourage research proposals from other federal agencies or their FFRDCs. They are eligible for NSF support only if a proposed project meets one or more of the exceptions spelled out below.
Under exceptional circumstances, research or science education projects at other federal agencies or their FFRDCs that can make unique contributions to the research needs of scientists elsewhere or to specific NSF objectives may receive NSF support. In particular, when contemplating establishment of new multi-user facilities, the NSF will consider using, expanding, or upgrading existing facilities of other agencies.
The NSF may fund research and logistic support activities of other government agencies or their FFRDCs directed at meeting the goals of special national and international research programs for which NSF bears special responsibility. Among such programs are the United States Antarctic Research Program, and cooperative science and technology programs with other countries.
Research and science education proposals submitted on behalf of faculty members at the military service academies, including the U.S. Military Academy, the U.S. Naval Academy, the U.S. Air Force Academy, the U.S. Coast Guard Academy, the U.S. Naval Postgraduate School, and the Uniformed University of the Health Sciences, will be accepted and subjected to competitive review and evaluation in the same way as proposals from other academic institutions. NSF will not ordinarily make an award to a service academy for major capital facilities and will make no award for routine functions of a service academy that normally would be funded from its own appropriations.
A scientist who has a joint appointment with a university and a federal agency, such as a Veterans Administration Hospital, or with a university and an FFRDC, may submit proposals through the university and may receive support if he or she is a bona fide faculty member of the university, though part of his or her salary may be provided by the Federal agency.
The CyberInfrastructure Partnership is funded by the National Science Foundation and includes the following partners: NCSA and SDSC.