Supercomputers/Systems/Niagara
Canada · decommissioned · hpc
Niagara
Operated by SciNet at SciNet (University of Toronto) .
Phase 1 seed dataset, compiled by hand. These rows were built from public operator, laboratory and vendor sources. A mechanical second-reader pass has since fetched every cited source: 1092 of 1153 systems have a readable citation that names them, and 40 are genuinely weakly sourced. Every claim carries its source and a confidence tier. Treat anything below verified as a lead, not a citation.
Measured performance
- Rmax
- 3.60 PFlop/s
- Rpeak
- 6.25 PFlop/s
- Rmax ÷ Rpeak
- 57.6%
- Cores
- 80,640
- Power
- 919.42 kW
- Per watt
- 3.9 GF/W
Figures are as last publicly reported for the configuration described below, not a live measurement. Where a system was upgraded in place, the post-upgrade configuration is shown and the earlier one appears in the timeline.
What this machine is made of
One row per supplier relationship. “Supplier at build” is the company that shipped the part at the time; where that company has since been acquired, the parent it rolls up to today is shown beside it. That distinction is what makes ticker-level aggregation possible across a thirty-year dataset.
| Role | Part | Supplier at build | Quantity | Confidence | Source |
|---|---|---|---|---|---|
| Integrator | - Lenovo ThinkSystem SD530 | Lenovo | - | Reported | top500.orgscinethpc.ca |
| CPU | Intel Xeon Gold 6148 CPUs x86-64 · 20C · 2.4 GHz · 14 nm · 150 W Xeon Gold 6148 20C 2.4GHz, two per node across 2,016 nodes after the 2020 expansion | Intel | - | Verified | top500.orgscinethpc.ca |
| Interconnect | Mellanox InfiniBand EDR Interconnect Fat tree · 100 Gb/s per 4x port · About 0.9 microseconds EDR InfiniBand in a Dragonfly+ topology | NVIDIA | - | Verified | top500.orgarxiv.org |
| Operating system | CentOS Operating systems Linux distribution, RHEL-compatible rebuild · Open source, community-maintained CentOS Linux 7 | The CentOS Project | - | Reported | top500.org |
The read
SciNet's second national system, a Lenovo ThinkSystem SD530 cluster owned by the University of Toronto and run within the Digital Research Alliance of Canada (formerly Compute Canada) federation. It launched on 5 March 2018 at 60,000 cores across 1,500 nodes of two Xeon Gold 6148 CPUs each, debuting 53rd on the June 2018 TOP500 at 3.07 PFlop/s. A 2020 expansion to 2,016 nodes and 80,640 cores raised it to the 3.60 PFlop/s Rmax and 6.25 PFlop/s Rpeak recorded here, first listed at 70th in June 2020; those are the figures used since they describe the machine for most of its life. TOP500's own field names the CPU as Xeon Gold 6148, though its page title oddly says 6248, so the field is trusted. The interconnect is EDR InfiniBand in a Dragonfly+ topology per SciNet's own paper on the system, and TOP500 independently confirms EDR. SciNet replaced it with Trillium on 7 August 2025; no source states an exact Niagara shutdown date, so the month of Trillium's launch is used.
Timeline
- 2018-03 First operational source
- 2020-06 List debut at #70 (June 2020) source
- 2025-08 Decommissioned source
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
Change history
- 2026-09-23 Added with a rank pointer on the edition matching its post-2020-expansion configuration
Source check
We fetched this system's own citations and recorded whether each page actually mentions it. This is a corroboration signal, not a fact check, and it is published so you can see how well the sourcing holds up rather than take it on trust.
| Cited source | Result | Found on the page |
|---|---|---|
| scinethpc.ca | names system and part | Mellanox InfiniBand EDR |
| top500.org | names system and part | Intel Xeon Gold 6148, Mellanox InfiniBand EDR, CentOS |
Checked 2026-10-09 by pnpm verify. Re-run it and the table changes with the web.
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
Inside a node
Latency-optimised versus throughput-optimised processors, NUMA, and the end of the host-to-device copy.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.