Resource Allocation Policy
Introduction and Overview
This document describes how access to the TAMU High Performance Research Computing (HPRC) resources. The computer systems managed by HPRC are available for use to university faculty, staff, and students who require large-scale computing capabilities. Access to and use of such systems is permitted only for academic research and instructional activity. All other use is strictly prohibited without the approval of the HPRC Director. Please see the Acceptable Use Policy for more information.
The policy is created under the supervision of the High Performance Research Computing Resource Allocation Committee (HPRC-RAC).
Terms defined in the Glossary are in italics the first time they appear.
Service Unit (SU)
A Service Unit (SU) is equivalent to the consumption of one hour (wall_time) on one cpu, where wall_time is your job finish time minus your job start time. The wall_time of a job is NOT the same as 1 hour of CPU time. The details of calculating SUs consumed by jobs are discussed here: AMS Service Unit.
Users need to submit applications for allocation of SUs on HPRC clusters at the Account Applications page . Once an application from a user is approved, the user can use those granted SUs to submit jobs on the cluster requested in the application. The SUs allocated to the user are the maximum SUs which can be consumed by all jobs belonging to the user.
Fiscal Year (FY)
A Fiscal Year (FY) starts on September 1st, and ends on August 31 next year. For example, FY 2017 starts on September 1, 2016, and ends on August 31, 2017. In general, the SUs allocated to a user are valid only during the Fiscal Year (FY) for which the SUs are allocated for and expire on August 31.
A Principal Investigator (PI) is a person who leads a research project, or a person who teaches an academic course.
For the purpose of HPRC allocations, only active faculty members and permanent research staff (subject to HPRC-RAC Chair review and approval) of Texas A&M System Members headquartered in Brazos County can serve as a PI. Adjunct and Visiting professors do not qualify themselves, but can use HPRC resources as part of a sponsoring PI’s group.
- A PI can have more than one allocation.
- A user can work on more than one project and with more than one PI
Allocations are made by granting Service Units (SUs) for projects to Principal Investigators (PIs). SUs are used on the computing resources by users associated with projects by PIs.
Types of Allocations
There are various types of SU allocations: Basic, Startup, Research, Sub-Allocation, and Other. Applications are required for Basic, Startup, and Research allocations. Table 1 shows the details for those three major types of allocations available to PIs, which are based on ranges of requested SUs.
|Allocation Type||Who can apply?||Minimum SUs per Allocation per Machine||Maximum SUs per Allocation per Machine||Maximum Total SUs per Machine||Maximum Number of Allocations per Machine||Allowed to spend more than allocation?||Reviewed and approved by|
|200,000||8,000,000||8,000,000||Determined by HPRC-RAC||No||HPRC-RAC|
|200,000||5,000,000||5,000,000||Determined by HPRC-RAC||No||HPRC-RAC|
Note: '*' - requires a PI
In addition, 20% of the common SU pool is reserved for special case assignments by the Director and the VPR.
Basic allocations are automatically created for each new user. Users can apply for accounts with basic allocations at any time, subject to approval of their PI. Basic allocations are approved by HPRC staff.
Startup allocations are intended to support low intensity projects, such as small scale analyses used on teaching, or benchmarking to assess the SU needs for a future research allocation. PIs can have a maximum of 2 concurrent startup allocations and can apply for them at any time. Startup allocations are approved by the HPRC director.
Research allocations are for medium to high intensity projects. PIs can have more than one research allocation, but are limited by the per-machine SU cap per academic year. Please refer to Table 1 - Details of Major Allocation Types for per-machine SU cap information.
If a PI reaches their SU cap, they and their users will continue having access to the supercomputing resources through a low priority queue. The low priority queue may be preemptible and will not use more than 15% of the total supercomputing resources at any one time. In addition, users of the low priority queue are limited to having no more than 50 jobs in such queue, with no more than 500 cores being used simultaneously.
Research allocations are to be reviewed monthly (during the second week of the month) by the HPRC-RAC.
Sub-Allocation for Project Members
A Sub-Allocation is the transfer of SUs from a PI's project account (Startup or Research) to a project member associated to that PI's project account. A sub-allocation does not require an application. Instead, a PI can make a sub-allocation at AMS, or just email such a request of sub-allocation to email@example.com.
A project member can receive a sub-allocation only if the user has an approved Basic or Startup account.
In addition to the three main categories, the HPRC can provide special case exceptions granted by the Director or the VPR. These are not intended to be for increasing the allocation of established project, but are for things like working with HPRC staff on new capabilities of general value to research communities across campus, new faculty startup, or other operations that go beyond normal research projects.
PIs who have contributed to the HPRC infrastructure via the “condo” contributions will also have committed allocations related to their contributions which are outside the common pool.
Allocation Availability on Machines
Research allocations are not available on small machines. Table 2 shows the allocation availability on our machines.
|Machines||Basic Allocation||Startup Allocation||Research Allocation|
A project account is number of 12 digits, e.g., 080123456789, and is used by HPRC users to submit jobs on machines. A project on a machine cannot be used on another machine. Allocations are allocated to project accounts. There are different types of project accounts, corresponding to the types of allocations.
Basic Project Account
A Basic Project Account, or a Basic Account, is a project account with SUs from Basic allocations.
Startup Project Account
A Startup Project Account, or a Startup Account, is a project account with SUs from Startup allocations. Normally, a Startup account may receive SUs from multiple Startup allocations.
Research Project Account
A Research Project Account, or a Research Account, is a project account with SUs from Research allocations. Normally, a Startup account may receive SUs from multiple Research allocations.
Sub-allocation Project Account
A Sub-allocation Project Account, or a Sub-allocation Account, is a project account with SUs from sub-allocations. A Sub-allocation account is created when a researcher is added to a Startup or a Research account after the PI of the Startup or research account emails such a request to firstname.lastname@example.org. A Sub-allocation account is associated to a Startup or a Research account.
A Sub-allocation account may receive SUs from multiple sub-allocations from its associated Startup or Research account. SUs can be transferred between a Sub-allocation account and its associated Startup or Research account. A PI can transfer SUs between the PI's Startup or Research account and the Sub-allocation accounts associated to that Startup or Research account at AMS, or email email@example.com to request such transfers of SUs with researcher name, net ID, and number of SUs to transfer.
Application and Review
All allocation applications are submitted through the HPRC Applications page. Startup and Research allocations should include text that explains:
- The purpose of the allocation,
- The time requested on the selected resource(s), including a calculation justifying the number of SUs requested. If the project is intended to benchmark the resources needed for a new project (startup only), rough approximations can be used if these are generated in consultation with HPRC staff.
- The intended methodology
- The state of the codes or applications to be used
Applications for Research Allocations must also include a PDF project description with the following components.
- Problem Statement – Section limited to 1 page, or less, describing the desired outcomes of the project.
- Background – Section limited to 1 page, or less, describing how the resources will be used to address the problem (i.e. student access for course work, specific models, etc).
- Methodology – Section limited to 1 page, or less, describing the computational methodology that will be used. This should include the applications required.
- Research Plan – Section limited to 1 page, or less, describing the research schedule, including the anticipated expenditure of granted resources. Allocations are assumed to be uniformly consumed over their lifetime. If this will not be the case, an estimate of expenditure by quarters is required.
- Requirements Analysis - Section limited to 2 pages, or less, detailing the basis for the requested computer time. Large allocations must exhibit an understanding of application efficiency, scaling, and provide accurate estimations of the time requirements.
The total length of the project description should be 5 pages or less, excluding any required reports summarizing results of prior support (required for renewals), references, and attachments.
Application for Basic and Startup allocations can be submitted at any time. Applications for Research Allocations should be submitted within the first week of each month.
The HPRC-RAC reviews all proposals for allocations exceeding 200K SUs/academic year. The merit criteria to evaluate each proposal is based on the scientific rationale for the proposed work, the proposed methodology and research plan, the careful analysis of the computational needs as well as prior (if relevant) performance, including actual usage, acknowledgement of the High Performance Research Computing (HPRC) group in publications and other vehicles for scientific dissemination.
PIs are strongly advised to carefully budget their usage appropriately throughout the year. Allocations are active until their expiration date, as long a positive resource balance remains. Once an allocation expires, or has been fully consumed, users accounts will be blocked from submitting work against the allocation. PIs should be prepared to submit new/renewal applications; there is not a mechanism for supplementing an expired allocation or one that has exceeded its SU limit.
The PI can monitor system usage and control authorization of project member accounts at the AMS portal. Automatic reminder emails will be sent by the management system as an allocation nears expiration. PIs are ultimately responsible for assuring that a current and actively monitored management email address has been assigned to each allocation.
User accounts must be associated with a valid allocation, and if not, will be retained for a maximum of 1 year pending authorization against a renewed or different allocation.
Acknowledgement of support
Any material, whether copyrighted or not, that reports findings obtained as a result of computation performed using TAMU’s HPRC resources should include an acknowledgement of support. Example acknowledgement:
“Portions of this research were conducted with high performance research computing resources provided by Texas A&M University (https://hprc.tamu.edu).
Failure to acknowledge HPRC support will negatively impact the outcome of future allocation proposals.
At the end of any Startup or Research allocation, a short summary report must be submitted to the allocation committee (see below for desired report content) at the AMS portal. Failure to submit this report may be used in the consideration of future allocations.
The report should be a maximum of 5 pages and should include the following information:
- Principal user information [PI] (name, status, department, phone, email, institution [if different from TAMU]
- Summarize the TAMU-sponsored research in a non-technical fashion, suitable for public consumption. (maximum two pages)
- Publications and other research products
- Describe any potential applications of the research to industry or government.
- Describe any use of this allocation to encourage education in computational science, and particularly the level of student involvement and any HPRC elements incorporated into formal courses.
For External Users
HPRC resources are available for use to external users who are actively collaborating with TAMU faculty or staff. HPRC supports TAMU researchers to stimulate the formation of collaborations with outside entities.
Accounts for users at not-for-profit external entities are not charged for usage, but may be added as sub-accounts under a TAMU faculty or staff Principal Investigator (PI).
Accounts for users at commercial for-profit entities are offered at commercial cost recovery rates (six cents per CPU core hour). TAMU faculty or staff who are collaborating with and supported by external for-profit entities are NOT subject to the commercial rate schedule, however, charges are made for usage by their external collaborators. Industry users will have to sign a Memorandum of Understanding (MoU) with HPRC to use HPRC resources. Users must apply for and receive their own accounts and may NOT be added as sub-accounts under a TAMU faculty or staff PI.
Determination of active collaboration with external entities and persons will be made annually, and when the TAMU collaborating PI asserts that a collaboration has ended. External users are expected to follow HPRC guidelines with respect to acknowledgement of support for their work.
NOTE: External users must provide their own licenses to use any licensed software on HPRC resources.
For Visiting Scholars, Affiliates, and Collaborators
At a bare minimum, you will need a TAMU assigned NetID. This provides login access to the TAMU campus and the HPRC clusters. The issuance of NetIDs is controlled by the TAMU Identity Management Office.
There are two types of NetIDs available for non-TAMU researchers.
- A Guest NetID, which will provide quick, but temporary, access to TAMU resources. This is best for short term use. Sponsoring professor required. A Guest NetID is good for short term consulting from vendors/suppliers and others. Long term collaborators should get a "supported" NetID.
- A "supported" NetID -- this requires a sponsor as well but provides a longer term login and access to TAMU wireless networks for those that visit the campus. In addition to a sponsor (PI), the Division of Research requires that collaborators using HPRC resources file a visiting scholars report. Users will need to apply for a Basic account to get started using HPRC resources. From there, their sponsoring PI can provide them "Service Units" (SUs=CPUs x hours) for use.
|Allocation||An amount of SUs assigned to a specific PI.|
|Condo||Researchers provide funds to purchase a large supercomputing cluster or data storage maintained in a centralized way by HPRC|
|Fiscal Year (FY)||A time period starts on September 1, and ends on August 31 next year.|
|HPRC||High Performance Research Computing at TAMU|
|Principal Investigator (PI)||A faculty member or research staff qualified to apply for allocations|
|Project Account||A 12 digits number used by users to submit jobs on machines.|
|Service Unit (SU)||The equivalent of 1 hour of wall clock time running on one processing core.|
|User||A user with a login account on one or more of the resources.|