Deep inside the construction of the Mathematics and Computer constructing (MC) on the University of Waterloo lies one of the refined and vital digital analysis computing programs in Canada. The newly refreshed system, named Nibi, just lately changed the Graham supercomputer.
The highly effective system has greater than 700 nodes and 140,000 CPU cores permitting Nibi to undertake numerous simultaneous computational jobs. It ranks in the top 500 research computing systems in the world.
“Nibi’s launch represents a defining moment for Canada’s digital research infrastructure,” said Mélanie Joly, Minister of Industry and Minister responsible for Canada Economic Development for Quebec Regions. “It equips our researchers with the advanced computing power they need to push the boundaries of discovery, strengthen our economy, and lead on the global stage. With its cutting-edge, energy-efficient design, Nibi is not only supporting today’s transformative projects, it is laying the foundation for the next generation of Canadian innovation from coast to coast to coast.”
“Nibi is the equivalent of 35,000 desktop computers working in harmony,” in line with John Morton (BMath ’94), who’s the director of Technology for SHARCNET and the technical lead for the Graham cluster. “Harnessing that power allows for ground-breaking research through batch computing, storage, visualization, GPU acceleration, cloud VMs, DBaaS and more being available to support Canadian research.”

While information centres are recognized to be enormous customers of energy, the designers of Nibi have sought to make use of modern technology to cut back cooling prices. The system makes use of water fairly than air to be extra environment friendly and sustainable. Instead of paying to eliminate warmth generated by high-performance computing, the warmth is captured and repurposed to warmth the Mike and Ophelia Lazaridis Quantum-Nano Centre.
To acknowledge this sustainable strategy, designers consulted with native Indigenous communities and determined to call the system Nibi. The identify means water in Anishinaabemowin (Ojibwe).
Nibi supplies GPU-equipped nodes every with eight H100 Nvidia GPUs to help giant AI fashions. With greater than 25 Petabytes of purely flash-based space for storing, Nibi supplies elevated efficiency and reliability from earlier storage options.

“Ontario is dwelling to a number of the brightest researchers in the world who’re driving Ontario’s key industries ahead,” mentioned Nolan Quinn, Minister of Colleges, Universities, Research Excellence and Security. “Through Compute Ontario, our government is proud to support the University of Waterloo’s innovative supercomputer, ensuring that researchers have the tools they need to make critical advancements in science, medicine, and technology, saving and improving lives across the province.”
Since 2017, Waterloo’s supercomputer has supported 1000’s of researchers from a number of universities to advance innovation throughout science, medicine and technology. As a member of SHARCNET, a multi-university consortium in Ontario, Nibi is projected to help greater than 4,000 researchers every year.
“At Waterloo, we’re proud to collaborate with our funding and research partners to advance the next wave of Canadian innovation,” says Charmaine Dean, vice-president, Research & International at Waterloo. “Nibi allows researchers across Canada the opportunity to explore and discover what’s possible, and we are pleased to continue Waterloo’s legacy as a computation leader.”

Waterloo has a rich history of computing dating back to its early years. Thanks in giant half to pioneering researchers equivalent to Don Cowan and Wes Graham, the establishment was the primary in Canada to accommodate probably the most highly effective pc of its time. The IBM 360/75 was the identical mannequin that NASA used to ship astronauts to the Moon.
Since then, Waterloo has continued to be a pacesetter in computing. To be sure that it stays on the cutting-edge, Waterloo researchers work carefully with quite a lot of companions. Nibi is funded by Innovation, Science and Economic Development by means of the Digital Research Alliance of Canada, and the Ministry of Colleges, Universities, Research Excellence and Security by means of Compute Ontario.