Supercomputers/Systems/Cirrus
United Kingdom · operational · hpc
Cirrus
Also known as Cirrus (HPE Cray EX4000)
Operated by EPCC .
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
- 4.12 PFlop/s
- Rpeak
- 4.97 PFlop/s
- Rmax ÷ Rpeak
- 82.9%
- Cores
- 70,560
- Power
- 319 kW
- Per watt
- 12.9 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 | - HPE Cray EX4000 | Hewlett Packard Enterprise | - | Reported | top500.orgepcc.ed.ac.uk |
| CPU | AMD EPYC 9005 (Turin) CPUs x86-64 · 192C · 2.7 GHz · TSMC N3E · 500 W AMD EPYC 9825 144C 2.2GHz, two per node on 256 nodes (arithmetic from node counts) | AMD | 512 estimated | Reported | top500.orgepcc.ed.ac.uk |
| Interconnect | HPE Slingshot-11 Interconnect Dragonfly · 200 Gb/s per port · Around 2 microseconds HPE Cray Slingshot 11, 200 Gbps per node | Hewlett Packard Enterprise | - | Reported | top500.orgdocs.cirrus.ac.uk |
| Storage | HPE Cray ClusterStor E1000 Storage Lustre appliance, flash and disk tiers · 700 PB on Frontier (Orion) · About 10 TB/s peak 1 PB HPE ClusterStor E1000 Lustre work file system, alongside 1 PB of Ceph | Hewlett Packard Enterprise | - | Reported | docs.cirrus.ac.uk |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription RHEL 9.5 per TOP500 | Red Hat | - | Reported | top500.org |
The read
EPCC's UKRI Tier 2 service after its hardware refresh, which replaced the earlier HPE SGI 8600 Cirrus with an HPE Cray EX4000 of dual AMD EPYC 9825 144-core nodes on Slingshot 11, running RHEL. EPCC dated the refreshed service to 11 December 2025 and counts 256 compute nodes (192 standard at 768 GB and 64 high-memory at 1,536 GB) and 73,728 cores; the Edinburgh International Data Facility page agrees, while the Cirrus documentation's hardware page still states 640 nodes and 184,320 cores, an inconsistency we could not resolve. TOP500 lists 70,560 cores, Rmax 4.12 PFlop/s, Rpeak 4.97 PFlop/s and 319 kW, ranked 314th in November 2025 and 343rd in June 2026. Storage is a 1 PB Ceph home file system and a 1 PB ClusterStor E1000 Lustre work file system. No GPU nodes are documented for the new system, so the older V100 partition is not carried over, and the physical hall is not stated in the fetched pages.
Timeline
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Change history
- 2026-10-02 Added Cirrus, EPCC's EPYC 9825 Tier 2 refresh, 314th on the November 2025 TOP500
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 |
|---|---|---|
| top500.org | names system and part | AMD EPYC 9005 (Turin), HPE Slingshot-11, HPE Cray ClusterStor E1000 |
| epcc.ed.ac.uk | names system and part | AMD EPYC 9005 (Turin), HPE Cray ClusterStor E1000 |
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.