The five sites to host the new national systems are:
- McGill University: The Béluga system is a heterogeneous multi-purpose supercomputer with four principal components: a CPU subsection, a GPU subsection, data-storage devices and a high-speed interconnect linking together the other three subsystems. Béluga was designed for a variety of workloads and is situated at ÉTS in Montreal. It is named for the St. Lawrence River’s Beluga whales. The CPU has 700 Intel servers, providing 28,000 compute cores of Intel Skylake generation. It’s comparable to 10,000 high-end laptops bundled into the same computer room.
- University of Victoria: The Arbutus system is an OpenStack cloud, with an emphasis on hosting virtual machines and other cloud workloads. Its powerful storage and compute capabilities are designed to support researchers processing, sharing and storing massive data sets. With 1.6 PB of storage available in a triple-redundant configuration and 7,640 CPU cores across 290 nodes, Arbutus can store the equivalent of 10 million eight-drawer filing cabinets worth of text and process calculations thousands of times faster than a desktop computer. The name Arbutus refers to the tree, which is commonly found on the south coast of Vancouver Island.
- Simon Fraser University: The Cedar system is a heterogeneous cluster, suitable for a variety of workloads. With over 3.6 petaFLOPS of computing power, Cedar has greater computational power than the entire fleet of Compute Canada’s aging legacy systems combined. It has large memory nodes — up to 3 terabytes — as well as on-node storage and GPU nodes and it’s designed to run multiple simultaneous jobs of up to 1280 CPU cores each. When launched, Cedar was Canada’s most powerful academic supercomputer and one of the world’s top 100 supercomputer installations.
- University of Waterloo: Named after Wes Graham, the first director of the Computer Centre at Waterloo, the Graham system is a heterogeneous cluster, suitable for a variety of workloads. The parallel file system and external persistent storage are similar to the Cedar System’s, but the interconnect is different and there is a slightly different mix of compute nodes. The Graham system is entirely liquid cooled, using rear-door heat exchangers. Graham was designed to support multiple simultaneous parallel jobs of up to 1,024 cores in a fully non-blocking manner. This system has a total of 41,548 cores and 520 GPU devices, spread across 1,185 nodes of different types.
- University of Toronto: The Niagara system was the 53rd fastest supercomputer on the Top500 list in June 2018 and is now at 76th. It is a homogenous cluster owned by the University of Toronto and run by SciNet and is intended to enable large parallel jobs of 1,040 cores and more. It was designed to optimize throughput of a range of scientific codes running at scale in an energy efficient way. It is a cluster of 2,016 Lenovo SD530 servers, each with 40 Intel Skylake at 2.4 GHz cores or 40 Intel CascadeLake cores at 2.5 GHz.
You can find more information on these national systems on our Compute Canada docs wiki, including helpful quick start guides.