This page provides additional information and resources to aid users submitting a Research Allocation Request.
The XSEDE training staff holds regularly scheduled "Writing and Submitting a Successful XSEDE Proposal" webinars timed with each quarterly XSEDE allocations cycle.
|Writing and Submitting a Successful XSEDE Proposal |
Recorded: March 22, 2017
The presentations detail all the necessary steps and documentation needed to submit a successful research allocation request. Topics covered in detail include:
- research objectives
- computational methodologies
- justification for SUs needed
- application efficiencies
The XSEDE Resource Allocation Committee (XRAC) typically awards larger allocations to requests that are well-written in terms of stating the need and objective of the supercomputing resource. Poorly written proposals lacking adequate detail, justification or organization may receive sharply reduced awards or be rejected outright. The XRAC reviewers do not evaluate the scientific merit of the proposal (unless there is no other funding specified) but they do evaluate the relationship of the research objectives to the computation resource.
Examples of well-written Research Requests, Startup Requests, Renewal Requests and Special Purpose Systems requests from a variety of domains are provided below. Please feel free to refer to these past requests when writing your own proposals.
When writing your research requests consider the following objectives and guidelines:
- Research must be clear and concise. Someone outside of the field should be able to understand the proposal. Why is the chosen technique preferred over another? Are the proposal's objectives well defined?
- Outline all the proposed computational algorithms and simulations
- Include any input parameters to be investigated and explain the choice of simulation parameters.
- Describe any previous results and progress. List all manuscripts (published, accepted, in preparation) made possible by using XSEDE resources. Clearly describe any progress to date. The XRAC tends to favorably review proposals that have a high "science-to-SU" ratio.
- Justify the choice of resources. Many research projects do not need access to high performance computing resources. Make a strong case both for why you require supercomputing time in general and why you want to use a particular resource. Valid justifications may include the need for any of the following:
- Intel KNL Co-Processors
- Large shared memory
- Large core counts for highly scalable applications
- Specialized software installed on XSEDE systems
- High performance parallel file systems
- Archival storage
- Low latency and/or high bandwidth interconnect
- Solid state (flash) scratch space
- GPGPUs Users requesting time on specialized systems such as Stampede, Gordon or visualization resources (Maverick) are expected to provide a stronger justification. Projects that involve running large numbers of independent jobs should consider using high throughput computing resources.
- Justify the amount of time requested. All proposals above the startup level are expected to contain a strong justification for the amount of time being requested, with larger requests generally being scrutinized more carefully. Provide detailed estimates of the computational needs based on the run times and scalability of the applications, problem sizes, the lengths of the simulations and the number of independent calculations. Break down your estimates by subproject and clearly summarize the results in a bulleted list or table. Consider using a Startup Allocation to obtain benchmark results in preparation for submission of a larger proposal.
Last update: March 22, 2017