Supercomputers/Systems/Cumulus Cascade Lake
United Kingdom · operational · hpc
Cumulus Cascade Lake
Also known as CSD3 Cascade Lake; cclake; Cumulus-Cascadelake
Operated by University of Cambridge at University of Cambridge (CSD3) .
Researched with AI assistance, checked by scripts. These rows were researched by AI agents reading public operator, laboratory and vendor sources, appended by scripts and checked by automated validation; no subject-matter expert has reviewed them (how this site is made). A mechanical second-reader pass has since fetched every cited source: 1350 of 1540 systems have a readable citation that names them, and 64 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
- 1.64 PFlop/s
- Rpeak
- 2.65 PFlop/s
- Rmax ÷ Rpeak
- 61.9%
- Cores
- 37,632
- Power
- -
- Per watt
- - 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 | - Dell PowerEdge C6420, built with StackHPC per the TOP500 entry title (CRC-StackHPC) | Dell Technologies | - | Reported | top500.org |
| CPU | - Intel Xeon Platinum 8276 28C 2.2GHz; 56 cores per node on 672 nodes implies two processors per node | Intel | 1,344 estimated | Reported | top500.orgdocs.hpc.cam.ac.uk |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds Mellanox InfiniBand HDR100 | NVIDIA | - | Verified | docs.hpc.cam.ac.uktop500.org |
| Operating system | Rocky Linux Operating systems RHEL-compatible Linux, community-maintained · Downstream rebuild of Red Hat Enterprise Linux source Rocky Linux 8 (TOP500 entry lists CentOS Scientific-OpenStack) | Rocky Enterprise Software Foundation | - | Reported | docs.hpc.cam.ac.uk |
The read
The Cascade Lake CPU partition of Cambridge's CSD3 service. The Research Computing Services documentation says the upgrade 'entered general service in November 2020', names the nodes cpu-p-[1-672], gives each cclake node 56 cores and 192 GiB (cclake-himem nodes 384 GiB), joins them with Mellanox HDR100 InfiniBand and now runs Rocky Linux 8. TOP500 lists the machine as Cumulus-Cascadelake, a Dell PowerEdge C6420 'CRC-StackHPC' system with Xeon Platinum 8276 28-core 2.2 GHz, HDR100 and CentOS Scientific-OpenStack, 37,632 cores (672 nodes of 56), Rmax 1.64 and Rpeak 2.65 PFlop/s and no power figure; it ranked 419th in November 2020 and 467th in June 2021, its last list. Documentation and TOP500 agree on the node arithmetic, the fabric and the start date, so the tier is verified, but the CPU model comes from TOP500 alone. The documentation page is a current CSD3 platform page, which is why status is held as of this check. TOP500 also lists the older Skylake and Xeon Phi Cumulus system separately; it is not recorded here. The GPU platform is Wilkes3 (wilkes-3).
Timeline
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
University of Cambridge (CSD3)
- Location
- Cambridge, United Kingdom
- Commissioned
- 2024
Change history
- 2026-10-18 Added Cumulus Cascade Lake
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.