News
EveryDatacenter
SciNet first wave installation of computer systems

SciNet

University of Toronto supercomputing consortium with clusters in a Vaughan warehouse

Photo: Sy maple, CC BY-SA 3.0, via Wikimedia Commons

SciNet is a high-performance computing consortium run by the University of Toronto together with affiliated Ontario hospitals, backed by federal and provincial money. Its IBM-built clusters went live in June 2009 and became the strongest university system of its kind outside the United States.

Share on WhatsApp

  • Ranked 53rd on the Top 500 list in November 2008 while still being built
  • GPC cluster: 30,240 Intel Nehalem cores in 3,780 IBM iDataPlex nodes
  • Water-cooled, with free-air mode once outdoor temperature falls low enough
  • Non-computing overhead kept under 20 percent of energy use
  • Backed by University of Toronto, four Ontario hospitals and government funds

What it is

SciNet is a computing consortium that pools high-performance resources for the University of Toronto and hospitals linked to it. Funding comes from Ottawa, the province, university units and those hospitals. The machines sit in a converted warehouse in Vaughan, roughly 30 km north of the St. George campus, because the campus could not supply the floor space and electricity needed. Two clusters make up the facility: a tightly coupled capability system and a much larger general purpose cluster. Together they handle jobs in quantum physics, weather, climate modelling, molecular structure work and particle collision analysis. In late 2008 the partly built hardware already sat at number 53 on the Top 500 ranking.

How it came about

Talks began in autumn 2004, when Canadian HPC groups agreed to answer the new National Platform Fund. Their joint bid succeeded and SciNet received a share of the money. A contract with IBM was signed in July 2008 and made public on 14 August that year. The main machine started running on 18 June 2009 and was described at the time as the country's fastest. Had it been finished half a year sooner it would have placed twelfth worldwide. The June 2009 Top 500 listing put the GPC iDataplex at 16 and the TCS at 80. Offices moved to the MaRS Discovery District in 2016.

The building and the halls

The machine hall covers 3,000 square feet (280 m2) and uses a raised floor. Cooling equipment includes a 735-ton chiller plus towers that support free-air operation. The GPC alone fills 45 racks of IBM System x iDataPlex dx360 M3 cabinets, each offering 84 rack units. Offices for staff are separate from the warehouse site, first on the St. George campus and later at MaRS.

Power and cooling

Electricity demand of the whole installation was compared to that of four thousand homes. Water carries heat away from the hardware. When outside air drops below a set temperature the plant switches the chiller off and pulls in cold ambient air instead, taking advantage of the Ontario winter. A typical facility spends about a third of its input on cooling and other overheads; this one was built to keep that share under 20 percent, a target set because climate studies were a central workload.

SciNet
A drawing made by EveryDatacenter from what this page says. It is a diagram, not a picture of the real data centre.

What runs inside

Work runs for university scientists and the partner hospitals, covering quantum mechanics, weather prediction, greenhouse warming scenarios, Arctic sea ice, and the shapes of chemicals, proteins, polymers and crystals. The facility also supports the ATLAS experiment and Large Hadron Collider work at CERN, examining collision data to probe why matter has mass and what the universe is made of.

The place

Vaughan, just north of Toronto, offered a warehouse with the room and electrical supply that a campus building could not provide. The cold local climate also helps, since outside air can be used for cooling for much of the year. Offices stayed closer to researchers, moving within Toronto in 2016.

Environment and impact

Efficiency was a design goal from the start, with the overhead share targeted below 20 percent against a norm near 33 percent. The consortium brings together a university, four hospitals and a cancer research institute, spreading both the cost and the scientific benefit across Ontario.

Network and links

Nodes talk over Gigabit Ethernet, and 864 of them add DDR InfiniBand for fast, low-latency message passing between parallel tasks. Links to CERN support the ATLAS and LHC programmes.

Questions people ask about SciNet

Who owns and runs SciNet?

It is a consortium led by the University of Toronto together with SickKids, Mount Sinai Hospital, the Ontario Institute for Cancer Research and Baycrest. Money comes from federal and provincial programmes, university units and the hospitals themselves.

Where is the hardware kept?

The clusters live in a warehouse in Vaughan, about 30 km north of the St. George campus, chosen for its floor area and power supply. Staff offices moved to the MaRS Discovery District in Toronto in 2016.

How is it cooled?

Water carries heat to a 735-ton chiller and cooling towers. A sensor watches outdoor temperature and, once it falls below a set point, the chiller steps aside so cold outside air does the work instead.

What kinds of research use it?

Climate and weather modelling, quantum physics, molecular structure calculations, and particle collision analysis for the ATLAS experiment at CERN, including work on why matter has mass.

How did it rank among supercomputers?

In November 2008 the half-finished system placed 53rd on the Top 500. After completion in June 2009 the GPC iDataplex reached 16th and the tightly coupled system sat at 80th.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "SciNet Consortium", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryDatacenter rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.

It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the data centre of the day, new pages on EveryDatacenter, and the conferences and anniversaries of the week ahead.