Available Resources

Compute Canada, in collaboration with many different partners from across Canada, including regional ARC-serving organizations and institutions, delivers world-class advanced research computing (ARC) infrastructure and support to Canadian researchers.

Each year, Compute Canada launches its annual Resource Allocation Competitions (RAC), a peer-reviewed process to grant priority access to storage and compute resources beyond what can be obtained via the Rapid Access Service.

Table of contents:

Resources available for RAC 2020


How to request resources

Resources available for RAC 2020

Below is the list of resources available for the 2020 Resource Allocation Competitions:

Region Cluster Sub-system (as shown on CCDB) Available resources for each sub-system Backup storage?
beluga-compute CPU No
beluga-gpu GPU No
ndc-calculquebec [Béluga storage] Project storage, Nearline storage Yes
VCPU, RAM, local ephemeral disk, volumes, snapshots, persistent cloud storage, floating IPs
beluga-persistent-cloud No
helios GPU No
helios-storage project storage No
Compute Ontario
graham-compute CPU No
graham-gpu GPU No
ndc-waterloo [Graham storage] Project storage, Nearline storage, dCache storage Yes
graham-persistent-cloud VCPU, RAM, local ephemeral disk, volumes, snapshots, persistent cloud storage, floating IPs No
niagara-compute CPU No
ndc-toronto [Niagara storage] Project storage Yes
HPSS Nearline storage No
HPC4Health-compute CPU No
HPC4Health-storage Project storage No
cedar-compute CPU No
cedar-gpu GPU No
ndc-sfu [Cedar storage] Project storage, Nearline storage, dCache storage Yes
VCPU, RAM, local ephemeral disk, volumes, snapshots, persistent cloud storage, floating IPs
VCPU, RAM, local ephemeral disk, volumes, snapshots, persistent cloud storage, floating IPs


In 2020, there will be a pilot project for the allocation of resources on a cluster provided by HPC4Health. Prior to the implementation of allocations in April 2020, users will be required to accept the following conditions with University Health Network (UHN), who will be acting on behalf of HPC4Health to onboard any projects:

1. Users agree not to act as a health information network provider under the Personal Health Information Protection Act (PHIPA), i.e., not to use the services to share the data to collaborators or through the web.

2. Users will sign a collaboration agreement with UHN.

3. Users will use two factor authentication to log in to the services.

Pre-consultation with HPC4Health is required in order for an allocation request to be considered. Contact our RAC team to arrange a consultation.

It is expected that up to 10 allocations will be awarded on the HPC4Health cluster. Access to this cluster via Rapid Access Service (RAS) is not planned. Because this is a pilot project, user groups are encouraged to consider how the HPC4Health resource may integrate with their own workflows and research programs. Service configuration will be as per Compute Canada standards (Slurm, CVMFS); by request, specific technical configurations could also be considered.  Availability and support is not yet fully determined and may change during the allocation year, but the platform is anticipated to be supported to the same degree as other host sites.

How to request resources in the online RAC application form

Note: There should be no discrepancies between what you request in the Application document attached to your application and what you request in the online form. In case of discrepancy, the online form will prevail.

  • In the corresponding Resources for Research Groups (RRG) Competition or Research Platforms and Portals (RPP) Competition online form on CCDB, go to the Resource Request section:
  • The dropdown menu under New resource request will show the list of all systems and subsystems (resources) available for allocation.
  • To indicate that your request can be allocated in any resource and that you do not have a preference for a particular one, please check the option “I have to select a system but I don’t mind receiving an allocation on any other suitable one” in the Reason for selecting system section.
  1. Requesting HPC resources: In the national clusters, compute and storage resources are shown as different subsystems on CCDB following this convention:
    • system-compute (e.g. cedar-compute): select these to request CPU resources. Please pay special attention to your memory requirements as this will be taken into account for final allocations.
    • system-gpu (e.g. cedar-gpu): select these to request GPU resources.
      • ndc-institution (e.g. ndc-sfu): select these to request storage resources on a particular system (for example, ndc-sfu is storage on Cedar; ndc-waterloo is storage on Graham). Each subsystem will only show the storage resources (/project, /nearline, etc.) available for that particular subsystem.
      • For example, if you want to request both CPU resources and /project storage resources on the Cedar cluster, you will have to fill in two separate requests: one with the cedar-compute subsystem for core years and memory, and another one with the ndc-sfu subsystem for /project storage (in TB).
  2. Requesting Cloud resources: If you need to request cloud resources in more than one location, then you will have to make one request for each location. On CCDB, cloud resources follow this convention:
    • system-compute-cloud or system-persistent-cloud: you can select compute or persistent VMs on Arbutus, Cedar and Béluga clouds, but only persistent VMs on graham-cloud. Each cloud resources will only show the specific “flavours” of VMs available for that particular site.

 Please email our RAC team with any questions on how to request resources.