RAC application guide

Note

Please read this guide carefully.

Submission dates¶

Application	Start	Deadline
* Resources for Research Groups (RRG) * Including Fast Track applications * Research Platforms and Portals (RPP) * Including RPP Annual Progress Reports	September 23, 2026	November 3, 2026, at 11:59 p.m. Eastern Standard Time

Warning

Extension of this deadline is not possible.

The results of the annual Resource Allocation Competition (RAC) will be announced in the spring.

For any questions, please contact allocations@tech.alliancecan.ca. We also recommend consulting the RAC Frequently Asked Questions page and Technical Glossary, as well as the annual report with the results of past RAC rounds (this includes the list of recipients with a summary of the awards).

Before you apply¶

We encourage all applicants to attend, or watch the recording of, one or several RAC information sessions this fall. The recordings will be available after the presentations here.
If you are applying for the first time, or you plan to submit a new application that will request resources that are asking for a Very Large amount of resources, we strongly encourage you to schedule a pre-submission consultation before October 30, 2026.

Tip

**A considerable amount of compute resources can be obtained through the Rapid Access Service (RAS) with a Default Resource Allocation Project (RAP) and without the need to apply for RAC**. Please consult the [Job Scheduling Policy](https://docs.alliancecan.ca/wiki/Job_scheduling_policies) and the [Allocation and compute scheduling](https://docs.alliancecan.ca/wiki/Allocations_and_compute_scheduling) documentation to better understand how to access compute resources without a RAC allocation—this can save you significant time and effort by avoiding submitting an unnecessary RAC application.

Your resource allocation is based on demonstrated need, not solely on previous year's usage. If your existing allocation does not meet your computational needs, you may request a minimal increase with proper justification. Conversely, if your usage is significantly lower than your existing allocation by the submission deadline, you must explain the reasons for the underutilization and the need to maintain the allocation. While a moderate underutilization does not automatically lead to the termination of an allocation, future allocations could be reduced or reverted to basic access to align with a historically low usage, particularly when no reasonable justification is provided.

Warning

Plan ahead and submit your application *before* the deadline: **late submissions will not be accepted.**

Warning

“Departmental" applications (that is, applications submitted on behalf of a group of PIs that may be from the same department or institution but are not collaborating in a common research project with clear goals and outcomes) *are not accepted*.

Warning

Do not submit multiple applications “just in case”: *all* applications submitted will be checked for eligibility but *only those that meet the eligibility requirements will be considered*: if you don't know what application pathway is best for you or have eligibility questions, please contact allocations@tech.alliancecan.ca *before* you start doing any work.

Terms of Access¶

By submitting a RAC application, you understand that

Resources awarded to successful applicants are provided FREE OF CHARGE.
Allocation decisions are the responsibility of the Resource Access Program Administrative Committee (RAPAC): these decisions are final and there is no appeal process. Applicants are welcome to share with the RAPAC any questions or concerns they may have about the results of an application and flag any potential errors. The RAPAC will analyze these concerns on a case-by-case basis and make every effort to answer questions, correct potential errors, and make reasonable accommodations when appropriate, but final allocation decisions remain at the discretion of the RAPAC.
All requests are scaled back every year based on different considerations that include, but are not limited to, the score from the peer review, the resource demand, capacity constraints, the size and computational expertise of the research group, and evidence of effective and efficient utilization of resources. Due to capacity constraints, there is no guarantee that a project will receive the same amount of resources than the previous year.
RAC allocations are not automatically renewed: an application must be submitted every year through any of the pathways available to confirm that the resources are still needed so that all requests can be properly reviewed. If no application is received, the existing allocation(s) will expire and no extension will be granted.
Only applications above the minimum passing score will be considered for an allocation. A successful score, however, does not guarantee that a project will receive the entirety of the resources requested.
The standard duration of a RAC allocation is one (1) year. Exceptionally, this duration may be extended or shortened during a specific competition round due to technical, logistical or other critical constraints (e.g. the delivery of new infrastructure is delayed).
Successful RAC applications are generally allowed to apply for continuation of resources for up to two years with the submission of a lightweight annual progress report; however, all requests received are reviewed and successful ones are allocated for one(1) year. For continuing projects, the score obtained from the original peer-review of the application will be carried over and used to make subsequent allocation decisions.
The PI is responsible for meeting the information requirements and overseeing the overall quality of the submitted RAC application, which will be evaluated based on the information provided.

Warning

Resources requested in the online form must match those listed in the application document and there should be no discrepancies between the two: *in case of discrepancy, what is requested in the online form will prevail.*

Any personal information collected by the Alliance is used only to review applications. Such information may be shared with relevant officials in the corresponding Alliance regional partner, as well as with the home institutions of the RAC applicant.
The Alliance will make publicly available the following information of successful projects: Name, Institution and Department of the PI; Project title and Summary; Field of research; Specific Research Area, and Amount allocated.
Researchers are required to acknowledge the Alliance and its regional partner organizations when publications or reports are published on research that uses the resources provided. For details about the wording to use, please visit Acknowledging the Alliance.

Application pathways¶

The RAC enables faculty members and their research groups to access compute, storage and cloud resources beyond what can be obtained via the RAS.

Note

These resources are provided FREE OF CHARGE.

	Resources for Research Groups (RRG)	Research Platforms and Portals (RPP)
Description	The RRG is a peer reviewed application process for projects whose primary purpose is to conduct research requiring compute, storage and cloud resources to meet their goals. Refer to the Appendices at the end of this Guide for the RRG Evaluation Criteria and scoring matrix. Projects primarily needing persistent instances in the cloud to provide a service through a platform or a portal should apply through the RPP application process instead.	The RPP is a peer reviewed application process for projects whose primary purpose is to provide a service to a larger research community through scientific gateways using a set of cloud-based tools or applications that improve access to shared datasets, enhance existing online research tools and facilities, or advance national or international research collaborations via a common interface. RPP projects must be able to develop, operate and manage the proposed portal or platform with minimal support from the Alliance. Please refer to the Appendices at the end of this Guide for the RPP Evaluation Criteria and the scoring matrix. Compute resources needed to support the backend of a RPP project can be requested with the application. However, projects primarily needing compute resources in a cluster to conduct research should apply through the RRG process instead.
Application document	The template provided is required. Word or OpenDocument (English, French) LaTeX (English, French) This link will open a non-editable Google document. Once there, go to File → Download in the main menu and select Microsoft Word or OpenDocument Format to download the template in the format of your choosing. Important: Maximum 10 pages (enforced). Only PDF format is accepted. Instructions may change from year to year so always use the latest version of the template.	The template provided is required. Word or OpenDocument (English, French) LaTeX (English, French) This link will open a non-editable Google document. Once there, go to File → Download in the main menu and select Microsoft Word or OpenDocument Format to download the template in the format of your choosing. Important: Maximum 10 pages (enforced). Only PDF format is accepted. Instructions may change from year to year so always use the latest version of the template.
Request period	RRG applications can request resources for one (1) year.	RPP applications can request resources for up to three (3) years in advance. However, resources are awarded for one (1) year and reassessed annually with the submission of a mandatory progress report.
Request for continuation	Users with an existing RRG award who meet the Fast Track eligibility are allowed to submit a lightweight progress report to renew their request for computational resources. Please read the Fast Track guideline carefully as this process includes important conditions and limitations that should be considered when deciding whether to use the Fast Track process or submit a new application. The Fast Track progress report must be completed online on CCDB.	Awarded multi-year projects do not need to submit a new application every year but are required to complete an annual progress report. The PI will be notified by email with instructions for the information required and the submission process. For more details, visit the RPP Progress Report page. However, if you were awarded a multi-year RPP that expires in April 2027, you must submit a new RPP application this fall if your need for resources will continue for the 2027-2028 allocation cycle. The RPP progress report must be completed online on CCDB.
Application limits	Eligible PIs can only submit one RRG application per contest (either a full application process or a Fast Track report). The same PI can simultaneously submit one RPP application during the same competition round. This PI can, however, be involved in multiple RRG or RPP submissions as a co-PI, as long as these collaborations are clearly disclosed in all the applications in which the PI is participating.	Eligible PIs may submit one RPP application per competition round, but can submit additional RPPs in subsequent competition rounds. The same PI can simultaneously submit one RRG or Fast Track application during the same competition round. This PI can, however, be involved in multiple RRG or RPP submissions as a co-PI, as long as these collaborations are clearly disclosed in all the applications in which the PI is participating.
Sub-projects	Each RRG application may include multiple sub-projects. While there is no limit to the number of sub-projects that can be included, we recommend a maximum of four (4) per application so that enough information about each sub-project can be provided within the page limit allowed.	Each RPP application must include only one (1) project. Sub-projects for RPP applications are not allowed.

Who can apply¶

RAC applicants must be a faculty member at a Canadian academic institution and have an Academic PI role (Faculty, Adjunct Faculty or Librarian). Users with Research Institute PI or Non-academic PI roles are not eligible to apply for RAC.

Important

First-time applicants may submit an application to the RAC after having used resources available through the Rapid Access Service (RAS) or by other means (e.g. in another country or at your institution) and determining, after consulting with a technical support staff, that these resources are no longer sufficient to meet their computational needs.
PIs must apply with their primary, most up-to-date position. If a PI recently moved to a different institution but has not yet applied for a new faculty role on the CCDB portal, please do so before submitting your RAC application. PIs with more than one active faculty role in the CCDB portal must ensure that the most up-to-date position is set as primary. To do so, go to the Home page on the CCDB portal and click on the Make this role primary button next to the role that you want to set as primary. Failing to do this could cause problems if your application is successful.

Contributors¶

Co-PI¶

A Co-PI is any faculty at a Canadian institution that is actively involved in the computational project. To be included in a RAC application, Co-PIs must have an active Academic PI role (Faculty, Adjunct Faculty or Librarian) on the CCDB portal. There is no limit to the number of co-PIs that can be listed in an application, but the role and involvement of each Co-PI must be clearly explained.

Delegated person¶

PIs can grant any other user with an active CCDB account permissions to complete and submit the RAC online form on their behalf. The PI or a senior member of the research group should review the application before it is submitted to make the entire document cohesive, to ensure that the eligibility and the information requirements are met, and to check that the application is complete: the PI is ultimately responsible for the quality of the proposal presented.

Secondary contact person¶

It is also possible to identify a person that our staff can contact to ask technical questions about your application. The Secondary contact person may be the same or different than the Delegated person, but it must have an active CCDB account. This person is expected to answer questions from our technical staff in a timely manner, ideally within 48 hours.

Collaborators¶

A collaborator is anyone actively involved in the project but that does not have a CCDB account. This may include researchers or students that are not based in Canada, as well as industry or government users. Collaborators can be mentioned in the Resource Management section of the Application Document and their involvement should be explained.

CV¶

The PI must upload an updated CV as part of the full application. CVs must use the harmonized, narrative style format CV template developed by the three federal granting agencies, the Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineering Research Council (NSERC), and the Social Sciences and Humanities Research Council (SSHRC). Please visit this page for detailed guidelines: https://cihr-irsc.gc.ca/e/53574.html

Important

When updating your CV, highlight what it is relevant to assess your computational research specifically (e.g., highlight publications enabled by the use of Alliance resources) and not your entire research program more broadly: this is what peer-reviewers evaluating your application will focus on.

Co-PIs are not required to submit a CV (Fast Track applications and RPP progress reports do not require submission of a new CV).

Minimum amount of resources eligible for RAC¶

Tip

If your need for computational resources can be met with what is available via the Rapid Access Service, do not submit a RAC application.

Warning

It is extremely important that the online application form includes a request for every resource (CPU, GPU, storage or cloud) that your project needs on each cluster. There should be NO DISCREPANCIES between the amount of resources requested in your application document and the online form. In case of discrepancy, what is requested in the online form will prevail.

Note that /home and /scratch storage are not allocated through the RAC.

Please visit the Available Resources page for a list of systems available for this competition and to understand how resources must be requested in the online form. If you don't understand some of the terms listed above, please visit the Technical Glossary.

HPC resources¶

To be eligible for RAC, your request for HPC resources must meet the specific conditions specified for each of the resources in the list below that you plan to request.

Resource	Unit	Minimum	Important information
CPU	Core-year or Core-equivalent-year	200	A CPU request must be equal to or greater than 200 core-years, or core-equivalent-years, in each of the clusters where these resources are needed. If you require 200 core-years or less but high memory per core, please use the formula below to calculate the core-equivalent-years. The resulting calculation must be equal to or greater than 200 in order to be eligible. `core-equivalent-years = MAX(core-years, mem requested / 4GB)` `Mem requested = core-years * mem per core requested` For more information about compute allocations, core-equivalent-years, and scheduler priority, please visit the Allocation and compute scheduling page.
GPU	RGU-year	25	A GPU request must be equal to or greater than 25 reference GPU units (RGU) years in each of the clusters where you need these resources. For more information about the GPU models available and what they will be charged for in RGU, visit the Allocation and compute scheduling page.
Project storage	TB	41	Contrary to CPU and GPU, where the minimum amount is required for each cluster, the minimum of 41 TB is the sum of all the project storage needed across all clusters. Important: This limit applies to General Purpose clusters only (Fir, Nibi, Rorqual and Narval). Any amount of project storage needed on Trillium must be requested through the RAC. If you need project and nearline storage, you can request both as long as one of them is equal to or greater* than the minimum eligible amount. In that case, the online form must include the project and nearline storage resources needed.
Nearline storage	TB	101	Contrary to CPU and GPU, where the minimum amount is required for each cluster, the minimum of 101 TB is the sum of all the nearline storage needed across all clusters. Important: This limit applies to General Purpose clusters only (Fir, Nibi, Rorqual and Narval) Any amount of project storage needed on HPSS must be requested through the RAC. If you need project and nearline storage, you can request both as long as one of them is equal to or greater* than the minimum eligible amount. In that case, the online form must include the project and nearline storage resources needed.
dCache storage*	-	-	Any amount of dCache storage must be requested.

Warning

dCache storage is primarily used by certain sub-atomic physics projects. Groups coming from a different field of research must consult with a technical staff before requesting this type of storage.

Examples of eligible HPC requests¶

If you need 180 core-years (or core-equivalent-years) and 50 TB of project storage on Fir, your RAC application should only request 50 TB of storage on that cluster. Do not request CPU in your application as you will be able to get the CPU that you need without a RAC allocation.
If you need 500 core-years, 20 TB of project storage and 80 TB of nearline storage on Rorqual, your RAC application should only request 500 core-years on that cluster, as your storage needs can be met with what is available via RAS.
If you need 100 core-years with 10 TB of project storage on Trillium and 50 TB of nearline storage on HPSS, then you should only request the 10 TB and 50 TB of storage on those clusters, respectively, as your need for CPU resources can be met without a RAC allocation.
If you need 400 core-years, 30 RGU-years and 200 TB of project storage on Nibi, then you must request all these resources in your RAC application as they are all above the minimum eligible amounts specified in the table.
If you need 150 core-years on Rorqual with different memory requirements: 100 cores with 16 GB per core (which results in 400 core-equivalent-years) and 50 cores with 4 GB per core (50 core-equivalent-years), then you should make two separate CPU requests on Rorqual to account for the different memory requirements. This request is acceptable because the total core-equivalent-years needed on a single cluster is greater than the minimum amount of CPU specified in the table. Another acceptable option would be to average out the total memory per core needed and just make one request for 150 cores with 12 GB per core.
If you need 45 TB of project storage and you want to split it between Rorqual and Narval as follows: 20 TB on Rorqual and 25 TB on Narval, then this request is eligible since the total amount of project storage needed exceeds 41 TB.
If you need a total of 100 TB of storage on Fir, split as follows: 50 TB of project storage and 50 TB of nearline, then this request is eligible because even though the nearline amount is below the minimum (101 TB), the project storage request is eligible. Therefore, you should include both requests in your application.

Examples of ineligible HPC requests¶

If you need 250 core-years in total but you want to split the CPU request between Fir and Nibi as follows: 100 cores with 4 GB per core on Fir and 150 cores with 4 GB per core on Nibi; this request will NOT be accepted because you should be able to get the CPU that you need without a RAC allocation. If your application only includes a CPU request in this way (that is, it does not include any additional storage, GPU or cloud resources), it will be rejected.
If you need 20 reference GPU units in total but you want to split the GPU request between Narval and Rorqual with 10 RGU and 10 RGU, respectively, this request will not be accepted.

Cloud Resources¶

Tip

If your need for cloud resources can be met with what is available via the Rapid Access Service, do not submit a RAC application.

To be eligible for RAC, your request for Cloud resources must be equal to or greater than at least one of the resources listed in the table below. The online form must include all cloud resources needed on each site requested, even if the amounts requested for some are below the minimum listed here. There is no minimum number of instances, volumes or snapshots required.

Warning

If you plan to request vGPUs in the cloud, you must justify in the Application Document why the project cannot use standard HPC GPUs.

As long as one of the resources needed in a particular cloud site is equal to or greater than the minimum specified in this table, the request is eligible. In that case, the online form must include all the cloud resources needed on that site, even if the amounts requested for some resources are below the minimum listed here.

Resource	Unit	Compute	Persistent
vCPU	vCPU-year	81	26
vGPU	RGU-year	3.6	N/A
RAM (GB)	GB	301	51
Floating IP	Number of IP	3
Volumes and Snapshot (persistent) storage	TB	11
Object storage	TB	11
Shared filesystem storage	TB	11

Assessing the size of your resource request¶

The table below provides applicants an idea of whether the amount of resources they plan to request is Small, Medium, Large or Very large by RAC standards.

Type	Resource	Unit	Small	Medium	Large	Very large
HPC	CPU	core-years	≤ 500	> 500 ≤ 2500	> 2500 ≤ 10000	> 10000
HPC	GPU	RGU-years	≤ 50	> 50 ≤ 250	> 250 ≤ 500	> 500
Cloud	vGPU	RGU-years	≤ 5	> 5 ≤ 10	> 10 ≤ 25	> 25
Cloud	Compute Cloud	vCPU-years	≤ 50	> 50 ≤ 250	> 250 ≤ 1000	> 1000
Cloud	Persistent Cloud	vCPU-years	≤ 25	> 25 ≤ 100	> 100 ≤ 250	> 250
HPC	Project storage	TB	≤ 50	> 50 ≤ 250	> 250 ≤ 1000	> 1000
HPC	Nearline storage	TB	≤ 250	> 250 ≤ 1000	> 1000 ≤ 2500	> 2500
Cloud	Volume and snapshot storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250
Cloud	Object storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250
Cloud	Shared filesystem storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250

Pre-submission consultations¶

While we encourage all applicants to consult with our technical staff if they have questions about eligibility or need help estimating your resource needs,

Pre-submission consultations are strongly recommended in the following cases:

The resource request of your RRG or RPP application falls into the Very Large category, as described in the Assessing the size of your resource request section.
You are applying for the first time: this includes PIs that are already using resources available through the RAS as well as new PIs, irrespective of whether they have previously used ARC resources.
Your project has very specific requirements such as large bursty usage needs (e.g., you would like to use a significant portion of a cluster).

The goal of the consultation is to

determine whether the resources needed for your project justify submitting a RAC application;
verify the project's eligibility to the right application process (RPP vs RRG);
provide technical assistance with the calculation of the resources needed;
determine the feasibility of meeting a very large request within our current resource constraints.

Tip

Consulting with a technical staff to prepare your application can help you build a stronger case for the resources needed—we ask you to mention in the application the name of the staff with whom you talked to and, when applicable, a brief summary with the results of that conversation (if you have a ticket number, please also mention it). Note, however, that participating in a pre-consultation does not guarantee the success of your application.

To allow adequate time for support by our technical staff, consultations must happen prior to October 30, 2026, so please request your consultation with enough time in advance. To schedule a consultation, contact allocations@tech.alliancecan.ca.

Evaluation process¶

Administrative review¶

Proposals not complying with the eligibility criteria will be rejected at this stage. During this phase, applicants may be contacted by an Alliance staff regarding potential errors.

Technical review¶

Eligible applications will receive one or several technical reviews. These reviewers are required to sign a Non-Disclosure Agreement prior to accessing any RAC application.

Warning

During this process, we may require additional information about your application and try to contact you (or the designated Secondary Contact Person) to ask some clarification questions. In order to ensure that an application can progress beyond the technical review, we expect you to respond in a timely manner (ideally within 48 hours). Failing to promptly communicate with our staff and provide clear answers to their questions may negatively impact your application.

The purpose of the technical review is to

ensure the appropriate system is requested by the PI and the required software is available;
evaluate application efficiency and scalability;
identify groups that may need help with application and workflow optimization;
identify discrepancies between the resource request included in the online form and the application document;
identify special software requirements;
provide a technical opinion on the reasonability of the request.

Peer review¶

Eligible applications are peer reviewed by interdisciplinary committees, which are responsible for evaluating and scoring applications and making recommendations about allocations. However, final decisions are made by the Resource Access Program Administrative Committee (RAPAC).

Many disciplinary experts volunteer each year to serve in a peer-review committee. All these experts are based in Canada and are required to accept the Alliance's Non-Disclosure Agreement and Conflict of Interest Policy.

While PIs have the option to select the committee that best matches the field(s) of research of their RAC project, applications can be moved to a different committee at the discretion of the committee Chairs.

The score resulting from this process is based on

the feasibility of the computational research project based on what research will be done and what outcomes will be delivered, and not on why it is important;
the appropriateness of the resources requested to achieve the objectives of the computational research project based on the technical justification provided; and
the likelihood that the resources requested will be efficiently used.

Full applications will be evaluated by one of the following committees:

Astronomy, Astrophysics and Cosmology
Bioinformatics
Chemistry, Biochemistry and Biophysics
Computer Sciences and Mathematics
Engineering
Environmental and Earth Sciences
Humanities and Social Sciences
Nano, Materials and Condensed Matter
Neurosciences, Medical Imaging and Medical Physics
Subatomic Physics, Nuclear Physics and Space Physics

Preparing your application¶

Please keep in mind the following when writing your RRG or RPP application:

Once completed, make sure that it is actually submitted: applications that are complete but not submitted will not be evaluated.

Warning

The online application must include a request *for every resource (CPU, GPU, storage or cloud) that your project needs on each cluster according to the [minimum eligible amounts](#minimum-amount-of-resources-eligible-for-rac)*. **There should be NO DISCREPANCIES between the amount of resources requested in your application document and what you request in the online form. *In case of discrepancy, what was requested in the online form will prevail*.**

Use the latest version of the application template provided and avoid writing an extremely technical proposal as the RAC peer-review committees are interdisciplinary.
References, when applicable, must be submitted as a separate document. However, if the total number of pages of the entire application, including References, is 10 pages or less, then submitting everything in just one document is acceptable.
If you are requesting CPU or GPU resources, you must use this Compute Calculator spreadsheet and include the table with the results of the calculations in your proposal. Note that while you can use decimals in this spreadsheet, the online form on CCDB only accepts integers, so you may round up the request to the next integer when completing the online request.
When available, CPU and GPU utilization information will be shown in the application form on CCDB, but the most accurate way to monitor utilization of your group is by going to https://portal.alliancecan.ca/slurm. Documentation about how to use this portal and read the charts is available on this page.
In general, RAC applicants find it difficult to estimate their GPU needs, which in most cases are over-requested and underutilized. We strongly encourage future RAC applicants to do two things before applying: start using GPU resources in order to get a better understanding of your needs, and consult with a technical support staff before submitting an application.
In case of GPU codes, you must discuss the GPU utilization that a typical job would achieve. Any choice of MIG or full GPUs must be justified with a scalability test performed with MIGs of different sizes and full GPUs. If a CPU version of the code exists, you must compare the performance of GPU and CPU versions. !!! warning Failing to provide this information may negatively impact your application and any potential GPU allocation. If you need assistance, contact allocations@tech.alliancecan.ca.
- Note that on Rorqual and Narval, GPU utilization data is available on their cluster job portals and can be used to test H100 and A100 GPU utilization, respectively (GPU utilization tests can be performed on Rorqual even if you are requesting H100 GPUs on other clusters).

Justification of underutilization¶

If the projected utilization of an allocation you hold is 50% or less, you must explain the reasons for the underutilization and describe any mitigation in place.

Important

We want to understand why resources allocated from the RAC pool to your research group may have not been fully utilized and what you plan to do about it. While we understand that there may be legitimate reasons for underutilization of resources allocated, we need to know what those reasons are to make appropriate allocation decisions—failing to provide a reasonable justification with sufficient details may negatively impact your application. Keep in mind the following:

The justification is only required for resources allocated in RAC Resource Allocation Projects (RAP), which are labelled rrg-name or rpp-name (for HPC allocations) and crg-name or cpp-name (for cloud allocations). For example, if you were allocated 500 core-years in your rrg-name RAP and you are projected to use 100 core-years in rrg-name but 300 core-years in def-name project, then you must justify the underutilization (100 core-years) of the 500 core-years allocated in the rrg-name project.
If the utilization of CPU or GPU resources in your Default project (def-name) is greater than the utilization in your RAC project (rrg-name or rpp-name), you must explain why this is the case and describe how your research group plans to address this issue to make effective and efficient use of your allocation.
If you have access to other computational resources (e.g. contributed hardware, access to a local cluster at your institution, etc.) within Canada or abroad, please also mention it.

How allocations are determined¶

Resources provided by the Alliance are limited, and for this reason requests for allocations are scaled every year based on the overall score of the application and the supply and demand. Allocation decisions follow the principle that the challenges arising from the shortage of resources and other constraints within the system are shared among all applicants.

The score resulting from the peer-review evaluation is used as the main reference for making allocation decisions. It is important to note, however, that the overall score does not in itself determine the amount of resources that a project will receive—other important considerations are also taken into account. These include, but are not limited to: resource demand, capacity constraints, evidence of effective and efficient utilization of resources, justification for the underutilization of an allocation, and the size of the research group. Allocation decisions are also impacted by variability in demand and supply from one system to another and from one resource type to another (e.g., GPU is in very high demand and short supply but CPU resources are less constrained).

The table below provides an overview of the approach followed to make allocation decisions.

| Resource | Scaling |Application Document (maximum 10 pages, PDF format only) - Maximum 10 pages (enforced). - Only PDF format is accepted. - Instructions may change from year to year so always use the latest version of the template. - [[RAC application guide#Appendix D: Non-disclosure Agreement and Conflict of Interest Policy for Peer Reviewers|Non-Disclosure Agreement and Conflict of Interest Policy]] should be [Non-Disclosure Agreement and Conflict of Interest Policy](#appendix-d-non-disclosure-agreement-and-conflict-of-interest-policy-for-peer-reviewers) as it's an internal link within the same document to an anchor. My earlier conversion was correct for this.

Double check the admonitions. - !!! note - green/blue for general info, good. - !!! warning - yellow/orange for important warnings, good. - !!! tip - blue for helpful tips, good. - !!! important - red for critical info, good.

Looks complete and accurate based on the instructions.!!! note Please read this guide carefully.

Submission dates¶

Application	Start	Deadline
* Resources for Research Groups (RRG) * Including Fast Track applications * Research Platforms and Portals (RPP) * Including RPP Annual Progress Reports	September 23, 2026	November 3, 2026, at 11:59 p.m. Eastern Standard Time

Warning

Extension of this deadline is not possible.

The results of the annual Resource Allocation Competition (RAC) will be announced in the spring.

For any questions, please contact allocations@tech.alliancecan.ca. We also recommend consulting the RAC Frequently Asked Questions page and Technical Glossary, as well as the annual report with the results of past RAC rounds (this includes the list of recipients with a summary of the awards).

Before you apply¶

We encourage all applicants to attend, or watch the recording of, one or several RAC information sessions this fall. The recordings will be available after the presentations here.
If you are applying for the first time, or you plan to submit a new application that will request resources that are asking for a Very Large amount of resources, we strongly encourage you to schedule a pre-submission consultation before October 30, 2026.

Tip

**A considerable amount of compute resources can be obtained through the Rapid Access Service (RAS) with a Default Resource Allocation Project (RAP) and without the need to apply for RAC**. Please consult the [Job Scheduling Policy](https://docs.alliancecan.ca/wiki/Job_scheduling_policies) and the [Allocation and compute scheduling](https://docs.alliancecan.ca/wiki/Allocations_and_compute_scheduling) documentation to better understand how to access compute resources without a RAC allocation—this can save you significant time and effort by avoiding submitting an unnecessary RAC application.

Your resource allocation is based on demonstrated need, not solely on previous year's usage. If your existing allocation does not meet your computational needs, you may request a minimal increase with proper justification. Conversely, if your usage is significantly lower than your existing allocation by the submission deadline, you must explain the reasons for the underutilization and the need to maintain the allocation. While a moderate underutilization does not automatically lead to the termination of an allocation, future allocations could be reduced or reverted to basic access to align with a historically low usage, particularly when no reasonable justification is provided.

Warning

Plan ahead and submit your application *before* the deadline: **late submissions will not be accepted.**

Warning

“Departmental" applications (that is, applications submitted on behalf of a group of PIs that may be from the same department or institution but are not collaborating in a common research project with clear goals and outcomes) *are not accepted*.

Warning

Do not submit multiple applications “just in case”: *all* applications submitted will be checked for eligibility but *only those that meet the eligibility requirements will be considered*: if you don't know what application pathway is best for you or have eligibility questions, please contact allocations@tech.alliancecan.ca *before* you start doing any work.

Terms of Access¶

By submitting a RAC application, you understand that

Resources awarded to successful applicants are provided FREE OF CHARGE.
Allocation decisions are the responsibility of the Resource Access Program Administrative Committee (RAPAC): these decisions are final and there is no appeal process. Applicants are welcome to share with the RAPAC any questions or concerns they may have about the results of an application and flag any potential errors. The RAPAC will analyze these concerns on a case-by-case basis and make every effort to answer questions, correct potential errors, and make reasonable accommodations when appropriate, but final allocation decisions remain at the discretion of the RAPAC.
All requests are scaled back every year based on different considerations that include, but are not limited to, the score from the peer review, the resource demand, capacity constraints, the size and computational expertise of the research group, and evidence of effective and efficient utilization of resources. Due to capacity constraints, there is no guarantee that a project will receive the same amount of resources than the previous year.
RAC allocations are not automatically renewed: an application must be submitted every year through any of the pathways available to confirm that the resources are still needed so that all requests can be properly reviewed. If no application is received, the existing allocation(s) will expire and no extension will be granted.
Only applications above the minimum passing score will be considered for an allocation. A successful score, however, does not guarantee that a project will receive the entirety of the resources requested.
The standard duration of a RAC allocation is one (1) year. Exceptionally, this duration may be extended or shortened during a specific competition round due to technical, logistical or other critical constraints (e.g. the delivery of new infrastructure is delayed).
Successful RAC applications are generally allowed to apply for continuation of resources for up to two years with the submission of a lightweight annual progress report; however, all requests received are reviewed and successful ones are allocated for one(1) year. For continuing projects, the score obtained from the original peer-review of the application will be carried over and used to make subsequent allocation decisions.
The PI is responsible for meeting the information requirements and overseeing the overall quality of the submitted RAC application, which will be evaluated based on the information provided.

Warning

Resources requested in the online form must match those listed in the application document and there should be no discrepancies between the two: *in case of discrepancy, what is requested in the online form will prevail.*

Any personal information collected by the Alliance is used only to review applications. Such information may be shared with relevant officials in the corresponding Alliance regional partner, as well as with the home institutions of the RAC applicant.
The Alliance will make publicly available the following information of successful projects: Name, Institution and Department of the PI; Project title and Summary; Field of research; Specific Research Area, and Amount allocated.
Researchers are required to acknowledge the Alliance and its regional partner organizations when publications or reports are published on research that uses the resources provided. For details about the wording to use, please visit Acknowledging the Alliance.

Application pathways¶

The RAC enables faculty members and their research groups to access compute, storage and cloud resources beyond what can be obtained via the RAS.

Note

These resources are provided FREE OF CHARGE.

	Resources for Research Groups (RRG)	Research Platforms and Portals (RPP)
Description	The RRG is a peer reviewed application process for projects whose primary purpose is to conduct research requiring compute, storage and cloud resources to meet their goals. Refer to the Appendices at the end of this Guide for the RRG Evaluation Criteria and scoring matrix. Projects primarily needing persistent instances in the cloud to provide a service through a platform or a portal should apply through the RPP application process instead.	The RPP is a peer reviewed application process for projects whose primary purpose is to provide a service to a larger research community through scientific gateways using a set of cloud-based tools or applications that improve access to shared datasets, enhance existing online research tools and facilities, or advance national or international research collaborations via a common interface. RPP projects must be able to develop, operate and manage the proposed portal or platform with minimal support from the Alliance. Please refer to the Appendices at the end of this Guide for the RPP Evaluation Criteria and the scoring matrix. Compute resources needed to support the backend of a RPP project can be requested with the application. However, projects primarily needing compute resources in a cluster to conduct research should apply through the RRG process instead.
Application document	The template provided is required. Word or OpenDocument (English, French) LaTeX (English, French) This link will open a non-editable Google document. Once there, go to File → Download in the main menu and select Microsoft Word or OpenDocument Format to download the template in the format of your choosing. Important: Maximum 10 pages (enforced). Only PDF format is accepted. Instructions may change from year to year so always use the latest version of the template.	The template provided is required. Word or OpenDocument (English, French) LaTeX (English, French) This link will open a non-editable Google document. Once there, go to File → Download in the main menu and select Microsoft Word or OpenDocument Format to download the template in the format of your choosing. Important: Maximum 10 pages (enforced). Only PDF format is accepted. Instructions may change from year to year so always use the latest version of the template.
Request period	RRG applications can request resources for one (1) year.	RPP applications can request resources for up to three (3) years in advance. However, resources are awarded for one (1) year and reassessed annually with the submission of a mandatory progress report.
Request for continuation	Users with an existing RRG award who meet the Fast Track eligibility are allowed to submit a lightweight progress report to renew their request for computational resources. Please read the Fast Track guideline carefully as this process includes important conditions and limitations that should be considered when deciding whether to use the Fast Track process or submit a new application. The Fast Track progress report must be completed online on CCDB.	Awarded multi-year projects do not need to submit a new application every year but are required to complete an annual progress report. The PI will be notified by email with instructions for the information required and the submission process. For more details, visit the RPP Progress Report page. However, if you were awarded a multi-year RPP that expires in April 2027, you must submit a new RPP application this fall if your need for resources will continue for the 2027-2028 allocation cycle. The RPP progress report must be completed online on CCDB.
Application limits	Eligible PIs can only submit one RRG application per contest (either a full application process or a Fast Track report). The same PI can simultaneously submit one RPP application during the same competition round. This PI can, however, be involved in multiple RRG or RPP submissions as a co-PI, as long as these collaborations are clearly disclosed in all the applications in which the PI is participating.	Eligible PIs may submit one RPP application per competition round, but can submit additional RPPs in subsequent competition rounds. The same PI can simultaneously submit one RRG or Fast Track application during the same competition round. This PI can, however, be involved in multiple RRG or RPP submissions as a co-PI, as long as these collaborations are clearly disclosed in all the applications in which the PI is participating.
Sub-projects	Each RRG application may include multiple sub-projects. While there is no limit to the number of sub-projects that can be included, we recommend a maximum of four (4) per application so that enough information about each sub-project can be provided within the page limit allowed.	Each RPP application must include only one (1) project. Sub-projects for RPP applications are not allowed.

Who can apply¶

RAC applicants must be a faculty member at a Canadian academic institution and have an Academic PI role (Faculty, Adjunct Faculty or Librarian). Users with Research Institute PI or Non-academic PI roles are not eligible to apply for RAC.

Important

First-time applicants may submit an application to the RAC after having used resources available through the Rapid Access Service (RAS) or by other means (e.g. in another country or at your institution) and determining, after consulting with a technical support staff, that these resources are no longer sufficient to meet their computational needs.
PIs must apply with their primary, most up-to-date position. If a PI recently moved to a different institution but has not yet applied for a new faculty role on the CCDB portal, please do so before submitting your RAC application. PIs with more than one active faculty role in the CCDB portal must ensure that the most up-to-date position is set as primary. To do so, go to the Home page on the CCDB portal and click on the Make this role primary button next to the role that you want to set as primary. Failing to do this could cause problems if your application is successful.

Contributors¶

Co-PI¶

A Co-PI is any faculty at a Canadian institution that is actively involved in the computational project. To be included in a RAC application, Co-PIs must have an active Academic PI role (Faculty, Adjunct Faculty or Librarian) on the CCDB portal. There is no limit to the number of co-PIs that can be listed in an application, but the role and involvement of each Co-PI must be clearly explained.

Delegated person¶

PIs can grant any other user with an active CCDB account permissions to complete and submit the RAC online form on their behalf. The PI or a senior member of the research group should review the application before it is submitted to make the entire document cohesive, to ensure that the eligibility and the information requirements are met, and to check that the application is complete: the PI is ultimately responsible for the quality of the proposal presented.

Secondary contact person¶

It is also possible to identify a person that our staff can contact to ask technical questions about your application. The Secondary contact person may be the same or different than the Delegated person, but it must have an active CCDB account. This person is expected to answer questions from our technical staff in a timely manner, ideally within 48 hours.

Collaborators¶

A collaborator is anyone actively involved in the project but that does not have a CCDB account. This may include researchers or students that are not based in Canada, as well as industry or government users. Collaborators can be mentioned in the Resource Management section of the Application Document and their involvement should be explained.

CV¶

The PI must upload an updated CV as part of the full application. CVs must use the harmonized, narrative style format CV template developed by the three federal granting agencies, the Canadian Institutes of Health Research (CIHR), the Natural Sciences and Engineering Research Council (NSERC), and the Social Sciences and Humanities Research Council (SSHRC). Please visit this page for detailed guidelines: https://cihr-irsc.gc.ca/e/53574.html

Important

When updating your CV, highlight what it is relevant to assess your computational research specifically (e.g., highlight publications enabled by the use of Alliance resources) and not your entire research program more broadly: this is what peer-reviewers evaluating your application will focus on.

Co-PIs are not required to submit a CV (Fast Track applications and RPP progress reports do not require submission of a new CV).

Minimum amount of resources eligible for RAC¶

Tip

If your need for computational resources can be met with what is available via the Rapid Access Service, do not submit a RAC application.

Warning

It is extremely important that the online application form includes a request for every resource (CPU, GPU, storage or cloud) that your project needs on each cluster. There should be NO DISCREPANCIES between the amount of resources requested in your application document and the online form. In case of discrepancy, what is requested in the online form will prevail.

Note that /home and /scratch storage are not allocated through the RAC.

Please visit the Available Resources page for a list of systems available for this competition and to understand how resources must be requested in the online form. If you don't understand some of the terms listed above, please visit the Technical Glossary.

HPC resources¶

To be eligible for RAC, your request for HPC resources must meet the specific conditions specified for each of the resources in the list below that you plan to request.

Resource	Unit	Minimum	Important information
CPU	Core-year or Core-equivalent-year	200	A CPU request must be equal to or greater than 200 core-years, or core-equivalent-years, in each of the clusters where these resources are needed. If you require 200 core-years or less but high memory per core, please use the formula below to calculate the core-equivalent-years. The resulting calculation must be equal to or greater than 200 in order to be eligible. `core-equivalent-years = MAX(core-years, mem requested / 4GB)` `Mem requested = core-years * mem per core requested` For more information about compute allocations, core-equivalent-years, and scheduler priority, please visit the Allocation and compute scheduling page.
GPU	RGU-year	25	A GPU request must be equal to or greater than 25 reference GPU units (RGU) years in each of the clusters where you need these resources. For more information about the GPU models available and what they will be charged for in RGU, visit the Allocation and compute scheduling page.
Project storage	TB	41	Contrary to CPU and GPU, where the minimum amount is required for each cluster, the minimum of 41 TB is the sum of all the project storage needed across all clusters. Important: This limit applies to General Purpose clusters only (Fir, Nibi, Rorqual and Narval). Any amount of project storage needed on Trillium must be requested through the RAC. If you need project and nearline storage, you can request both as long as one of them is equal to or greater* than the minimum eligible amount. In that case, the online form must include the project and nearline storage resources needed.
Nearline storage	TB	101	Contrary to CPU and GPU, where the minimum amount is required for each cluster, the minimum of 101 TB is the sum of all the nearline storage needed across all clusters. Important: This limit applies to General Purpose clusters only (Fir, Nibi, Rorqual and Narval) Any amount of project storage needed on HPSS must be requested through the RAC. If you need project and nearline storage, you can request both as long as one of them is equal to or greater* than the minimum eligible amount. In that case, the online form must include the project and nearline storage resources needed.
dCache storage*	-	-	Any amount of dCache storage must be requested.

Warning

dCache storage is primarily used by certain sub-atomic physics projects. Groups coming from a different field of research must consult with a technical staff before requesting this type of storage.

Examples of eligible HPC requests¶

If you need 180 core-years (or core-equivalent-years) and 50 TB of project storage on Fir, your RAC application should only request 50 TB of storage on that cluster. Do not request CPU in your application as you will be able to get the CPU that you need without a RAC allocation.
If you need 500 core-years, 20 TB of project storage and 80 TB of nearline storage on Rorqual, your RAC application should only request 500 core-years on that cluster, as your storage needs can be met with what is available via RAS.
If you need 100 core-years with 10 TB of project storage on Trillium and 50 TB of nearline storage on HPSS, then you should only request the 10 TB and 50 TB of storage on those clusters, respectively, as your need for CPU resources can be met without a RAC allocation.
If you need 400 core-years, 30 RGU-years and 200 TB of project storage on Nibi, then you must request all these resources in your RAC application as they are all above the minimum eligible amounts specified in the table.
If you need 150 core-years on Rorqual with different memory requirements: 100 cores with 16 GB per core (which results in 400 core-equivalent-years) and 50 cores with 4 GB per core (50 core-equivalent-years), then you should make two separate CPU requests on Rorqual to account for the different memory requirements. This request is acceptable because the total core-equivalent-years needed on a single cluster is greater than the minimum amount of CPU specified in the table. Another acceptable option would be to average out the total memory per core needed and just make one request for 150 cores with 12 GB per core.
If you need 45 TB of project storage and you want to split it between Rorqual and Narval as follows: 20 TB on Rorqual and 25 TB on Narval, then this request is eligible since the total amount of project storage needed exceeds 41 TB.
If you need a total of 100 TB of storage on Fir, split as follows: 50 TB of project storage and 50 TB of nearline, then this request is eligible because even though the nearline amount is below the minimum (101 TB), the project storage request is eligible. Therefore, you should include both requests in your application.

Examples of ineligible HPC requests¶

If you need 250 core-years in total but you want to split the CPU request between Fir and Nibi as follows: 100 cores with 4 GB per core on Fir and 150 cores with 4 GB per core on Nibi; this request will NOT be accepted because you should be able to get the CPU that you need without a RAC allocation. If your application only includes a CPU request in this way (that is, it does not include any additional storage, GPU or cloud resources), it will be rejected.
If you need 20 reference GPU units in total but you want to split the GPU request between Narval and Rorqual with 10 RGU and 10 RGU, respectively, this request will not be accepted.

Cloud Resources¶

Tip

If your need for cloud resources can be met with what is available via the Rapid Access Service, do not submit a RAC application.

To be eligible for RAC, your request for Cloud resources must be equal to or greater than at least one of the resources listed in the table below. The online form must include all cloud resources needed on each site requested, even if the amounts requested for some are below the minimum listed here. There is no minimum number of instances, volumes or snapshots required.

Warning

If you plan to request vGPUs in the cloud, you must justify in the Application Document why the project cannot use standard HPC GPUs.

As long as one of the resources needed in a particular cloud site is equal to or greater than the minimum specified in this table, the request is eligible. In that case, the online form must include all the cloud resources needed on that site, even if the amounts requested for some resources are below the minimum listed here.

Resource	Unit	Compute	Persistent
vCPU	vCPU-year	81	26
vGPU	RGU-year	3.6	N/A
RAM (GB)	GB	301	51
Floating IP	Number of IP	3
Volumes and Snapshot (persistent) storage	TB	11
Object storage	TB	11
Shared filesystem storage	TB	11

Assessing the size of your resource request¶

The table below provides applicants an idea of whether the amount of resources they plan to request is Small, Medium, Large or Very large by RAC standards.

Type	Resource	Unit	Small	Medium	Large	Very large
HPC	CPU	core-years	≤ 500	> 500 ≤ 2500	> 2500 ≤ 10000	> 10000
HPC	GPU	RGU-years	≤ 50	> 50 ≤ 250	> 250 ≤ 500	> 500
Cloud	vGPU	RGU-years	≤ 5	> 5 ≤ 10	> 10 ≤ 25	> 25
Cloud	Compute Cloud	vCPU-years	≤ 50	> 50 ≤ 250	> 250 ≤ 1000	> 1000
Cloud	Persistent Cloud	vCPU-years	≤ 25	> 25 ≤ 100	> 100 ≤ 250	> 250
HPC	Project storage	TB	≤ 50	> 50 ≤ 250	> 250 ≤ 1000	> 1000
HPC	Nearline storage	TB	≤ 250	> 250 ≤ 1000	> 1000 ≤ 2500	> 2500
Cloud	Volume and snapshot storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250
Cloud	Object storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250
Cloud	Shared filesystem storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250

Pre-submission consultations¶

While we encourage all applicants to consult with our technical staff if they have questions about eligibility or need help estimating your resource needs,

Pre-submission consultations are strongly recommended in the following cases:

The resource request of your RRG or RPP application falls into the Very Large category, as described in the Assessing the size of your resource request section.
You are applying for the first time: this includes PIs that are already using resources available through the RAS as well as new PIs, irrespective of whether they have previously used ARC resources.
Your project has very specific requirements such as large bursty usage needs (e.g., you would like to use a significant portion of a cluster).

The goal of the consultation is to

determine whether the resources needed for your project justify submitting a RAC application;
verify the project's eligibility to the right application process (RPP vs RRG);
provide technical assistance with the calculation of the resources needed;
determine the feasibility of meeting a very large request within our current resource constraints.

Tip

Consulting with a technical staff to prepare your application can help you build a stronger case for the resources needed—we ask you to mention in the application the name of the staff with whom you talked to and, when applicable, a brief summary with the results of that conversation (if you have a ticket number, please also mention it). Note, however, that participating in a pre-consultation does not guarantee the success of your application.

To allow adequate time for support by our technical staff, consultations must happen prior to October 30, 2026, so please request your consultation with enough time in advance. To schedule a consultation, contact allocations@tech.alliancecan.ca.

Evaluation process¶

Administrative review¶

Proposals not complying with the eligibility criteria will be rejected at this stage. During this phase, applicants may be contacted by an Alliance staff regarding potential errors.

Technical review¶

Eligible applications will receive one or several technical reviews. These reviewers are required to sign a Non-Disclosure Agreement prior to accessing any RAC application.

Warning

During this process, we may require additional information about your application and try to contact you (or the designated Secondary Contact Person) to ask some clarification questions. In order to ensure that an application can progress beyond the technical review, we expect you to respond in a timely manner (ideally within 48 hours). Failing to promptly communicate with our staff and provide clear answers to their questions may negatively impact your application.

The purpose of the technical review is to

ensure the appropriate system is requested by the PI and the required software is available;
evaluate application efficiency and scalability;
identify groups that may need help with application and workflow optimization;
identify discrepancies between the resource request included in the online form and the application document;
identify special software requirements;
provide a technical opinion on the reasonability of the request.

Peer review¶

Eligible applications are peer reviewed by interdisciplinary committees, which are responsible for evaluating and scoring applications and making recommendations about allocations. However, final decisions are made by the Resource Access Program Administrative Committee (RAPAC).

Many disciplinary experts volunteer each year to serve in a peer-review committee. All these experts are based in Canada and are required to accept the Alliance's Non-Disclosure Agreement and Conflict of Interest Policy.

While PIs have the option to select the committee that best matches the field(s) of research of their RAC project, applications can be moved to a different committee at the discretion of the committee Chairs.

The score resulting from this process is based on

the feasibility of the computational research project based on what research will be done and what outcomes will be delivered, and not on why it is important;
the appropriateness of the resources requested to achieve the objectives of the computational research project based on the technical justification provided; and
the likelihood that the resources requested will be efficiently used.

Full applications will be evaluated by one of the following committees:

Astronomy, Astrophysics and Cosmology
Bioinformatics
Chemistry, Biochemistry and Biophysics
Computer Sciences and Mathematics
Engineering
Environmental and Earth Sciences
Humanities and Social Sciences
Nano, Materials and Condensed Matter
Neurosciences, Medical Imaging and Medical Physics
Subatomic Physics, Nuclear Physics and Space Physics

Preparing your application¶

Please keep in mind the following when writing your RRG or RPP application:

Once completed, make sure that it is actually submitted: applications that are complete but not submitted will not be evaluated.

Warning

The online application must include a request *for every resource (CPU, GPU, storage or cloud) that your project needs on each cluster according to the [minimum eligible amounts](#minimum-amount-of-resources-eligible-for-rac)*. **There should be NO DISCREPANCIES between the amount of resources requested in your application document and what you request in the online form. *In case of discrepancy, what was requested in the online form will prevail*.**

Use the latest version of the application template provided and avoid writing an extremely technical proposal as the RAC peer-review committees are interdisciplinary.
References, when applicable, must be submitted as a separate document. However, if the total number of pages of the entire application, including References, is 10 pages or less, then submitting everything in just one document is acceptable.
If you are requesting CPU or GPU resources, you must use this Compute Calculator spreadsheet and include the table with the results of the calculations in your proposal. Note that while you can use decimals in this spreadsheet, the online form on CCDB only accepts integers, so you may round up the request to the next integer when completing the online request.
When available, CPU and GPU utilization information will be shown in the application form on CCDB, but the most accurate way to monitor utilization of your group is by going to https://portal.alliancecan.ca/slurm. Documentation about how to use this portal and read the charts is available on this page.
In general, RAC applicants find it difficult to estimate their GPU needs, which in most cases are over-requested and underutilized. We strongly encourage future RAC applicants to do two things before applying: start using GPU resources in order to get a better understanding of your needs, and consult with a technical support staff before submitting an application.
In case of GPU codes, you must discuss the GPU utilization that a typical job would achieve. Any choice of MIG or full GPUs must be justified with a scalability test performed with MIGs of different sizes and full GPUs. If a CPU version of the code exists, you must compare the performance of GPU and CPU versions. !!! warning Failing to provide this information may negatively impact your application and any potential GPU allocation. If you need assistance, contact allocations@tech.alliancecan.ca.
- Note that on Rorqual and Narval, GPU utilization data is available on their cluster job portals and can be used to test H100 and A100 GPU utilization, respectively (GPU utilization tests can be performed on Rorqual even if you are requesting H100 GPUs on other clusters).

Justification of underutilization¶

If the projected utilization of an allocation you hold is 50% or less, you must explain the reasons for the underutilization and describe any mitigation in place.

Important

We want to understand why resources allocated from the RAC pool to your research group may have not been fully utilized and what you plan to do about it. While we understand that there may be legitimate reasons for underutilization of resources allocated, we need to know what those reasons are to make appropriate allocation decisions—failing to provide a reasonable justification with sufficient details may negatively impact your application. Keep in mind the following:

The justification is only required for resources allocated in RAC Resource Allocation Projects (RAP), which are labelled rrg-name or rpp-name (for HPC allocations) and crg-name or cpp-name (for cloud allocations). For example, if you were allocated 500 core-years in your rrg-name RAP and you are projected to use 100 core-years in rrg-name but 300 core-years in def-name project, then you must justify the underutilization (100 core-years) of the 500 core-years allocated in the rrg-name project.
If the utilization of CPU or GPU resources in your Default project (def-name) is greater than the utilization in your RAC project (rrg-name or rpp-name), you must explain why this is the case and describe how your research group plans to address this issue to make effective and efficient use of your allocation.
If you have access to other computational resources (e.g. contributed hardware, access to a local cluster at your institution, etc.) within Canada or abroad, please also mention it.

How allocations are determined¶

Resources provided by the Alliance are limited, and for this reason requests for allocations are scaled every year based on the overall score of the application and the supply and demand. Allocation decisions follow the principle that the challenges arising from the shortage of resources and other constraints within the system are shared among all applicants.

The score resulting from the peer-review evaluation is used as the main reference for making allocation decisions. It is important to note, however, that the overall score does not in itself determine the amount of resources that a project will receive—other important considerations are also taken into account. These include, but are not limited to: resource demand, capacity constraints, evidence of effective and efficient utilization of resources, justification for the underutilization of an allocation, and the size of the research group. Allocation decisions are also impacted by variability in demand and supply from one system to another and from one resource type to another (e.g., GPU is in very high demand and short supply but CPU resources are less constrained).

The table below provides an overview of the approach followed to make allocation decisions.

| Resource | Scaling |Resource |Unit |Compute |Persistent | |:------------------------------------------|:--------------|:----------|:---------------| | vCPU | vCPU-year | 81 | 26 | | vGPU | RGU-year | 3.6 | N/A | | RAM (GB) | GB | 301 | 51 | | Floating IP | Number of IP | 3 | | | Volumes and Snapshot (persistent) storage | TB | 11 | | | Object storage | TB | 11 | | | Shared filesystem storage | TB | 11 | |

Assessing the size of your resource request¶

The table below provides applicants an idea of whether the amount of resources they plan to request is Small, Medium, Large or Very large by RAC standards.

Type	Resource	Unit	Small	Medium	Large	Very large
HPC	CPU	core-years	≤ 500	> 500 ≤ 2500	> 2500 ≤ 10000	> 10000
HPC	GPU	RGU-years	≤ 50	> 50 ≤ 250	> 250 ≤ 500	> 500
Cloud	vGPU	RGU-years	≤ 5	> 5 ≤ 10	> 10 ≤ 25	> 25
Cloud	Compute Cloud	vCPU-years	≤ 50	> 50 ≤ 250	> 250 ≤ 1000	> 1000
Cloud	Persistent Cloud	vCPU-years	≤ 25	> 25 ≤ 100	> 100 ≤ 250	> 250
HPC	Project storage	TB	≤ 50	> 50 ≤ 250	> 250 ≤ 1000	> 1000
HPC	Nearline storage	TB	≤ 250	> 250 ≤ 1000	> 1000 ≤ 2500	> 2500
Cloud	Volume and snapshot storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250
Cloud	Object storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250
Cloud	Shared filesystem storage	TB	≤ 10	> 10 ≤ 50	> 50 ≤ 250	> 250

Pre-submission consultations¶

While we encourage all applicants to consult with our technical staff if they have questions about eligibility or need help estimating your resource needs,

Pre-submission consultations are strongly recommended in the following cases:

The resource request of your RRG or RPP application falls into the Very Large category, as described in the Assessing the size of your resource request section.
You are applying for the first time: this includes PIs that are already using resources available through the RAS as well as new PIs, irrespective of whether they have previously used ARC resources.
Your project has very specific requirements such as large bursty usage needs (e.g., you would like to use a significant portion of a cluster).

The goal of the consultation is to

determine whether the resources needed for your project justify submitting a RAC application;
verify the project's eligibility to the right application process (RPP vs RRG);
provide technical assistance with the calculation of the resources needed;
determine the feasibility of meeting a very large request within our current resource constraints.

Tip

Consulting with a technical staff to prepare your application can help you build a stronger case for the resources needed—we ask you to mention in the application the name of the staff with whom you talked to and, when applicable, a brief summary with the results of that conversation (if you have a ticket number, please also mention it). Note, however, that participating in a pre-consultation does not guarantee the success of your application.

To allow adequate time for support by our technical staff, consultations must happen prior to October 30, 2026, so please request your consultation with enough time in advance. To schedule a consultation, contact allocations@tech.alliancecan.ca.

Evaluation process¶

Administrative review¶

Proposals not complying with the eligibility criteria will be rejected at this stage. During this phase, applicants may be contacted by an Alliance staff regarding potential errors.

Technical review¶

Eligible applications will receive one or several technical reviews. These reviewers are required to sign a Non-Disclosure Agreement prior to accessing any RAC application.

Warning

During this process, we may require additional information about your application and try to contact you (or the designated Secondary Contact Person) to ask some clarification questions. In order to ensure that an application can progress beyond the technical review, we expect you to respond in a timely manner (ideally within 48 hours). Failing to promptly communicate with our staff and provide clear answers to their questions may negatively impact your application.

The purpose of the technical review is to

ensure the appropriate system is requested by the PI and the required software is available;
evaluate application efficiency and scalability;
identify groups that may need help with application and workflow optimization;
identify discrepancies between the resource request included in the online form and the application document;
identify special software requirements;
provide a technical opinion on the reasonability of the request.

Peer review¶

Eligible applications are peer reviewed by interdisciplinary committees, which are responsible for evaluating and scoring applications and making recommendations about allocations. However, final decisions are made by the Resource Access Program Administrative Committee (RAPAC).

Many disciplinary experts volunteer each year to serve in a peer-review committee. All these experts are based in Canada and are required to accept the Alliance's Non-Disclosure Agreement and Conflict of Interest Policy.

While PIs have the option to select the committee that best matches the field(s) of research of their RAC project, applications can be moved to a different committee at the discretion of the committee Chairs.

The score resulting from this process is based on

the feasibility of the computational research project based on what research will be done and what outcomes will be delivered, and not on why it is important;
the appropriateness of the resources requested to achieve the objectives of the computational research project based on the technical justification provided; and
the likelihood that the resources requested will be efficiently used.

Full applications will be evaluated by one of the following committees:

Astronomy, Astrophysics and Cosmology
Bioinformatics
Chemistry, Biochemistry and Biophysics
Computer Sciences and Mathematics
Engineering
Environmental and Earth Sciences
Humanities and Social Sciences
Nano, Materials and Condensed Matter
Neurosciences, Medical Imaging and Medical Physics
Subatomic Physics, Nuclear Physics and Space Physics

Preparing your application¶

Please keep in mind the following when writing your RRG or RPP application:

Once completed, make sure that it is actually submitted: applications that are complete but not submitted will not be evaluated.

Warning

The online application must include a request *for every resource (CPU, GPU, storage or cloud) that your project needs on each cluster according to the [minimum eligible amounts](#minimum-amount-of-resources-eligible-for-rac)*. **There should be NO DISCREPANCIES between the amount of resources requested in your application document and what you request in the online form. *In case of discrepancy, what was requested in the online form will prevail*.**

Use the latest version of the application template provided and avoid writing an extremely technical proposal as the RAC peer-review committees are interdisciplinary.
References, when applicable, must be submitted as a separate document. However, if the total number of pages of the entire application, including References, is 10 pages or less, then submitting everything in just one document is acceptable.
If you are requesting CPU or GPU resources, you must use this Compute Calculator spreadsheet and include the table with the results of the calculations in your proposal. Note that while you can use decimals in this spreadsheet, the online form on CCDB only accepts integers, so you may round up the request to the next integer when completing the online request.
When available, CPU and GPU utilization information will be shown in the application form on CCDB, but the most accurate way to monitor utilization of your group is by going to https://portal.alliancecan.ca/slurm. Documentation about how to use this portal and read the charts is available on this page.
In general, RAC applicants find it difficult to estimate their GPU needs, which in most cases are over-requested and underutilized. We strongly encourage future RAC applicants to do two things before applying: start using GPU resources in order to get a better understanding of your needs, and consult with a technical support staff before submitting an application.
In case of GPU codes, you must discuss the GPU utilization that a typical job would achieve. Any choice of MIG or full GPUs must be justified with a scalability test performed with MIGs of different sizes and full GPUs. If a CPU version of the code exists, you must compare the performance of GPU and CPU versions. !!! warning Failing to provide this information may negatively impact your application and any potential GPU allocation. If you need assistance, contact allocations@tech.alliancecan.ca.
- Note that on Rorqual and Narval, GPU utilization data is available on their cluster job portals and can be used to test H100 and A100 GPU utilization, respectively (GPU utilization tests can be performed on Rorqual even if you are requesting H100 GPUs on other clusters).

Justification of underutilization¶

If the projected utilization of an allocation you hold is 50% or less, you must explain the reasons for the underutilization and describe any mitigation in place.

Important

We want to understand why resources allocated from the RAC pool to your research group may have not been fully utilized and what you plan to do about it. While we understand that there may be legitimate reasons for underutilization of resources allocated, we need to know what those reasons are to make appropriate allocation decisions—failing to provide a reasonable justification with sufficient details may negatively impact your application. Keep in mind the following:

The justification is only required for resources allocated in RAC Resource Allocation Projects (RAP), which are labelled rrg-name or rpp-name (for HPC allocations) and crg-name or cpp-name (for cloud allocations). For example, if you were allocated 500 core-years in your rrg-name RAP and you are projected to use 100 core-years in rrg-name but 300 core-years in def-name project, then you must justify the underutilization (100 core-years) of the 500 core-years allocated in the rrg-name project.
If the utilization of CPU or GPU resources in your Default project (def-name) is greater than the utilization in your RAC project (rrg-name or rpp-name), you must explain why this is the case and describe how your research group plans to address this issue to make effective and efficient use of your allocation.
If you have access to other computational resources (e.g. contributed hardware, access to a local cluster at your institution, etc.) within Canada or abroad, please also mention it.

How allocations are determined¶

Resources provided by the Alliance are limited, and for this reason requests for allocations are scaled every year based on the overall score of the application and the supply and demand. Allocation decisions follow the principle that the challenges arising from the shortage of resources and other constraints within the system are shared among all applicants.

The score resulting from the peer-review evaluation is used as the main reference for making allocation decisions. It is important to note, however, that the overall score does not in itself determine the amount of resources that a project will receive—other important considerations are also taken into account. These include, but are not limited to: resource demand, capacity constraints, evidence of effective and efficient utilization of resources, justification for the underutilization of an allocation, and the size of the research group. Allocation decisions are also impacted by variability in demand and supply from one system to another and from one resource type to another (e.g., GPU is in very high demand and short supply but CPU resources are less constrained).

The table below provides an overview of the approach followed to make allocation decisions.

| Resource | Scaling |Scaling | |:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------