As part of its strategy for Advanced Research Computing and High Performance Computing in Canada, Compute Canada has conducted a site selection for four new systems. These systems are intended to replace and augment the currently aging computational systems available to Canadian academic researchers.
Recognizing the diversity of ARC computing in academic research, Compute Canada is planning to install four systems. Three systems will be so-called General Purpose clusters, aimed at small to moderate sized jobs with a large variety of demands (e.g, IO, GPUS, memory, …).
The fourth machine will be a Large Parallel system, i.e. a tightly coupled parallel supercomputer intended for running large (on the order of at least 512 cores per job) parallel jobs, typically using the Message Passing Interface.
SciNet, at the University of Toronto, has been selected as the site for the LP system. The GP systems will be at the University of Victoria, at Simon Fraser University, and at the University of Waterloo.
Note that it is very hard at this stage to know when these new systems will be online. A rough, very tentative estimate is that they could start arriving sometime in 2016.
For more information regarding the selection, see the selection announcement by Compute Canada.