Supercomputers/Systems/COSMOS (LUNARC)
Sweden · operational · hpc
COSMOS (LUNARC)
Also known as COSMOS
Operated by LUNARC at Lund University (LUNARC) .
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 1581 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
- -
- 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 |
|---|---|---|---|---|---|
| Accelerator | - NVIDIA A100 on 6 AMD-side and 4 Intel-side nodes | NVIDIA | - | Verified | lunarc.lu.sesupr.naiss.se |
| Accelerator | - NVIDIA A40 on 6 AMD-side and 5 Intel-side nodes | NVIDIA | - | Verified | lunarc.lu.sesupr.naiss.se |
| CPU | - AMD EPYC 7413 (Milan), 2 per node on 182 nodes | AMD | - | Verified | lunarc.lu.sesupr.naiss.se |
| CPU | - Intel Cascade Lake CPU nodes, 22 nodes of 32 cores | Intel | - | Verified | lunarc.lu.sesupr.naiss.se |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds HDR InfiniBand 100 Gbit per node with 200 Gbit switches | NVIDIA | - | Reported | lunarc.lu.se |
| Storage | - IBM Spectrum Scale, 3 PB project storage | IBM | - | Reported | lunarc.lu.se |
The read
Lund University's main LUNARC cluster, installed in 2023 and the next entry after Aurora (2016, 300 nodes) in LUNARC's system history. LUNARC and the NAISS allocation portal agree on the layout: 182 AMD nodes funded by Lund University (two EPYC 7413 Milan, 48 cores, 256 GB, 2 TB local disk), 6 A100 and 6 A40 GPU nodes on the AMD side, 22 Intel Cascade Lake CPU nodes of 32 cores, and 5 A40 plus 4 A100 GPU nodes on the Intel side. LUNARC adds HDR InfiniBand at 100 Gbit per node, 3 PB of IBM Spectrum Scale, Rocky Linux 9 and Slurm. GPU cards per node are not stated, so no GPU quantity is recorded; vendor and peak performance are not published. The SUPR page shows COSMOS open for Lund-local allocation rounds in 2026, which is the basis for the status date.
Timeline
- 2023 First operational source
Site & facility
Change history
- 2026-10-29 Added COSMOS, Lund University's EPYC and A100 cluster at LUNARC
Further reading
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.
Storage and I/O
Parallel file systems, why checkpointing dominates the write load, and the metadata failure mode nobody expects.