Niagara at Scale Pilot
March 5, 2021 in blog-general, for_press, for_researchers, for_users, frontpage, news
SciNet will be reserving the Niagara cluster for two days in March for the first-ever “Niagara at Scale”, from March 30th, 2021, at 12 noon EST, to April 1st, 2021, at 12 noon EST.
Purpose of the “Niagara at Scale” event
This event will enable pre-approved projects that require all or nearly all of the capacity of the Niagara supercomputer at once. Such heroic computations are Niagara’s mandate, as it is the “Large Parallel cluster within the national systems of the Compute Canada Federation, and the fastest machine of its kind in Canada according to the TOP500 List. But computations of this size — think massively parallel codes running on tens of thousands of cores — are hard or impossible to run within the regular batch scheduler.
How to apply
We already have some groups interested in participating, but we would like to extend our invitation to the whole Canadian high-performance computing community before committing to a particular date. Users that have massively parallel jobs or workflows that could take advantage of this opportunity, are encourage to contact us at support@scinet.utoronto.ca by Friday, March 12, 2021 (note: this is an extension of the original deadline of March 5).
In the email, please briefly describe your intended computation, as well as the size and duration of the jobs you would like to run at scale. Successful proposals will need to show evidence that their codes can run efficiently on at least 20,000 cores on Niagara and include strong and/or weak scaling data and plots.
In addition, your codes must be able to checkpoint and restart, especially since jobs will be restricted to shorter wall time.
Information session on March 10, 2021
We will hold an online information session regarding this program on March 10, 2021 at our SciNet User Group Meeting at noon EST. Attend to learn what kind of computations this program is aimed at. We will also provide guidance on how to get your computation to such a large scale if it needs it but your code does not yet scale to that size. For more information and sign-up for the event, go to https://scinet.courses/569
Future “Niagara at Scale” Events
The current event is a pilot project. If this initiative proves successful, we are planning to hold several of these events per year.