Supercomputers/Systems/COSMA8
United Kingdom · operational · hpc
COSMA8
Also known as DiRAC Memory Intensive Durham
Operated by Durham University at Durham University COSMA Facility .
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
- -
- Rpeak
- -
- Rmax ÷ Rpeak
- -
- Cores
- -
- 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 EMC PowerEdge C6525 servers | Dell Technologies | - | Reported | nvidianews.nvidia.com |
| CPU | - 360 compute nodes with 2x AMD EPYC Rome processors and 1 TB RAM (quantity is nodes) | AMD | 360 reported | Reported | dirac.ac.uk |
| CPU | AMD EPYC 7763 (Milan) CPUs x86-64 · 64C · 2.45 GHz · 7 nm · 280 W 168 compute nodes with AMD EPYC Milan 7763 processors and 1 TB RAM (quantity is nodes) | AMD | 168 reported | Reported | dirac.ac.uk |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds Non-blocking HDR200 InfiniBand fat tree with in-network computing | NVIDIA | - | Verified | dirac.ac.uknvidianews.nvidia.com |
The read
The Durham University COSMA system that provides the DiRAC Memory Intensive service for large cosmological simulations. The DiRAC Durham page lists 528 compute nodes (360 with two AMD EPYC Rome processors and 168 with AMD EPYC Milan 7763), each with 1 TB of RAM, a non-blocking HDR200 InfiniBand network, 13 PB of bulk Lustre storage, 1.2 PB of fast scratch at about 350 GB/s, a 26 PB tape archive, two 4 TB fat nodes, a high-cache Milan-X node and 7 GPU nodes with 18 GPUs (A100, V100, MI200 and MI100). NVIDIA's April 2021 announcement confirms Dell EMC PowerEdge C6525 servers and a full non-blocking fat-tree HDR 200 Gb/s InfiniBand network with in-network computing. The system has not been submitted to TOP500 as far as the fetched sources show, so no FLOPS or core figure is recorded; the exact system-wide core count is left blank rather than estimated.
Site & facility
Change history
- 2026-10-01 Added COSMA8 (Durham University)
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 |
|---|---|---|
| dirac.ac.uk | names system, part and count | NVIDIA InfiniBand HDR · counts: 168 |
| nvidianews.nvidia.com | names system and part | AMD EPYC 7763 (Milan), NVIDIA InfiniBand HDR |
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.