The Compute Canada national merit based access is the envy of many research communities around the world because the policy allows access to academic, government and private sector researchers. Any grant-eligible faculty member at any Canadian university may obtain a Compute Canada account. The faculty member may sponsor any number of users under their account. These sponsored users may include students, research associates or external collaborators - including industry collaborators and academic collaborators outside of Canada.
All usage by any sponsored user is accounted for under the allocation of the sponsoring faculty member. The vast majority of current Compute Canada users are sponsored in this way by Canadian academic faculty and access the facilities free of charge.
Researchers who are not grant-eligible, for example, those from industry or government laboratories, may also create an account and sponsor others but must pay a fee to enable this access.
The number of access requests has been growing each year. The table below presents the number of active faculty accounts captured on January first each year for the last 6 years. In addition, it includes some relevant numbers for the annual allocation process, described below.
|Faculty accounts (beginning of year)||1,415||1,854||2,229||2,343||2,721||3,146|
|All accounts (beginning of year)||4,171||6,278||7,754||8,151||9,570||10,799|
|Number of RAC/RPP Requests||135||161||212||289||346||366|
|RAC Requests (1,000s of core years)||73||104||142||173||192||238|
|RAC Requests (storage in terabytes)||8,169||9,903||12,412||13,422||20,871||28,620|
swipe to view full table
The majority of Compute Canada compute resources (approximately 80%) on each system are allocated through a competitive peer-review process collectively referred to as Compute Canada’s Resource Allocation Competitions (RAC). This committee review ensures that excellent science guides the usage of CC facilities. RAC includes two streams: “regular RAC” and “Research Platforms and Portals (RPP)”.
RPP was introduced for the first time in 2014 and includes many projects serving shared datasets to large collaborations or to the research community at large.
The remaining 20% of available compute resources are left to the “default allocation”.
Significant science impact is achieved through default usage, enabled by a low barrier to access to researchers with modest needs. While the number of papers and citations obtained by the largest RAC users is substantially higher than a typical default user, the impact per paper, as measured by FWCI averaged over similar sized cohorts, is roughly the same.
In the RAC process, all applications are categorized by discipline and assigned to one of seven national disciplinary expert committees for scientific review. The science review process produces a set of scores for each proposal, with a scoring system modelled on NSERC procedures.
For the 2016 competition, 80 volunteer faculty members from across Canada served as reviewers for the RAC process. A national multi-disciplinary committee formed primarily of expert committee chairs then balances the “budget” between all proposals. In addition, Compute Canada team members with representation from all regions perform a detailed technical review to determine the reasonableness of each request and the optimal site on which to implement it.