Supercomputers/Systems/Pirineus III
Spain · operational · mixed
Pirineus III
Also known as CSUC Pirineus III
Operated by CSUC .
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
- 2.20 PFlop/s
- 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 | - Supermicro AS-1125HS-TNR general-purpose nodes | Supermicro | - | Reported | csuc.cat |
| Integrator | - Atos SBI-420P-1T3N blade nodes | Atos | - | Reported | csuc.cat |
| Integrator | - Dell PowerEdge GPU nodes with NVIDIA H100 | Dell Technologies | - | Reported | csuc.cat |
| Accelerator | - NVIDIA H100 80 GB, 2 per node on 29 GPU nodes (58 cards by arithmetic) | NVIDIA | 58 estimated | Verified | csuc.catres2.bsc.es |
| CPU | - AMD EPYC 9654 96-core 2.4 GHz, 2 per node on 58 nodes (50 standard with 384 GB, 8 high-memory with 3 TB) | AMD | - | Verified | csuc.catres2.bsc.es |
| CPU | - AMD EPYC 9334 32-core, 2 per node on 29 GPU nodes | AMD | - | Verified | csuc.catres2.bsc.es |
| CPU | - Intel Xeon Gold 6342 24-core 2.8 GHz, 2 per node on 40 older standard nodes with 256 GB | Intel | - | Reported | csuc.cat |
| Interconnect | NVIDIA InfiniBand HDR Interconnect Fat tree or dragonfly+ · 200 Gb/s per port · About 0.6 microseconds InfiniBand HDR 200 Gb/s | NVIDIA | - | Reported | csuc.cat |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Red Hat Enterprise Linux 8.9 | Red Hat | - | Reported | csuc.cat |
The read
Third generation of the Pirineus cluster run by the Catalan university consortium CSUC, installed in 2023 and, per CSUC, in operation from October 2024; it follows the Bull Sequana X800/X550 system of 2018 that ran until 2023. CSUC's 2025 spec sheet lists 50 standard nodes (two EPYC 9654, 384 GB), 8 high-memory nodes (same CPUs, 3 TB), 40 older nodes with two Xeon Gold 6342 (256 GB) and 29 GPU nodes with two EPYC 9334 and two H100 80 GB each, on InfiniBand HDR at 200 Gb/s under RHEL 8.9, and quotes 2.2 PFlop/s peak. The Spanish Supercomputing Network (RES) page, maintained by BSC, shows only the 58-node EPYC 9654 partition (11,136 cores, 427 TFlop/s) and the 29-node GPU partition (1.59 PFlop/s), about 2.0 PFlop/s combined, and says 34 percent of the capacity goes to RES; the two figures do not reconcile, so only CSUC's 2.2 is recorded and no core total is given. CSUC's history page names Supermicro, Atos and Dell as node suppliers, and CSUC pitches the GPU block at AI model work.
Timeline
- 2024-10 First operational source
Change history
- 2026-10-29 Added Pirineus III, CSUC's H100 and EPYC cluster in Barcelona
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.
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.