Supercomputers/Systems/Occigen2
France · decommissioned · hpc
Occigen2
Also known as Occigen 2; Occigen Haswell plus Broadwell configuration
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
- 2.49 PFlop/s
- Rpeak
- 3.57 PFlop/s
- Rmax ÷ Rpeak
- 69.7%
- Cores
- 85,824
- Power
- 1.43 MW
- Per watt
- 1.7 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.
How it compares
Ranked against the other machines in this dataset that report the same figure: those that first came into service within two years of Occigen2, and those in France. Computed from our tables; a metric is ranked only when at least eight peers report it.
| Metric | Value | Against its era | Against its country |
|---|---|---|---|
| Rmax | 2.49 PFlop/s | #77 of 234 · top 33% | #23 of 52 · top 43% |
| Energy efficiency | 1.7 GF/W | #123 of 158 · bottom 23% | #30 of 42 · bottom 30% |
| Rmax ÷ Rpeak | 69.7 % | #123 of 229 · bottom 47% | #40 of 51 · bottom 23% |
| Power draw | 1,430 kW | #32 of 161 · top 19% | #10 of 42 · top 22% |
Every metric is ranked largest-first, so a high power-draw rank means a big power draw, not a good one. Rmax ÷ Rpeak uses HPL-measured results only. Rankings move as machines are added; they describe this dataset, not the world.
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 | - Bull bullx DLC 720 | Bull | - | Reported | top500.orgcines.fr |
| CPU | Intel Xeon E5-2690 v3 CPUs x86-64 · 12C · 2.6 GHz · 22 nm · 135 W 2,106 Haswell nodes (12-core 2.6 GHz processors, two per node) | Intel | 2,106 reported | Reported | cines.fr |
| CPU | - 1,260 Broadwell nodes, Xeon E5-2690v4 14C 2.6GHz, two per node | Intel | 1,260 reported | Reported | cines.frtop500.org |
| Interconnect | Mellanox InfiniBand FDR Interconnect Fat tree · 56 Gb/s per 4x port · About 1 microsecond Mellanox InfiniBand 4x FDR, three-level fat tree | Mellanox NVDANASDAQ
rolls up to NVIDIA | - | Reported | cines.frtop500.org |
| Storage | Panasas ActiveScale (PanFS) Storage Object storage, blade shelves (1 director + 10 storage blades) 260 TB Panasas file system for user files, 10 GB/s | Panasas | - | Reported | cines.fr |
| Cooling | Direct liquid cooling, warm water Cooling Direct-to-chip cold plates, facility water loop · Removes 90+ percent of node heat to liquid Direct Liquid Cooling with warm water at 35 C | Bull | - | Reported | cines.fr |
The read
The full Occigen machine after its 2016-2017 extension, as TOP500 lists it separately from the original Haswell-only entry (occigen). CINES's activity report tabulates Occigen 1 (2,106 Haswell nodes, 50,544 cores, 2.1 PFlop/s peak) plus Occigen 2 (1,260 Broadwell nodes of 14-core 2.6 GHz processors, 35,280 cores, 1.46 PFlop/s), giving 3,366 nodes, 85,824 cores and 3.5 PFlop/s over 38 compute racks, warm-water (35 C) direct liquid cooling, InfiniBand FDR, and a 5.1 PB Lustre store at 106 GB/s; a CINES gazette of November 2017 adds 260 TB of Panasas for user files and says the extension entered production in early 2017 after a 'grands challenges' period (its own text gives 3366 and 3336 nodes in two places). TOP500 lists the same 85,824 cores at Rmax 2.49 and Rpeak 3.57 PFlop/s and 1,430 kW, with an installation year of 2016. The gazette puts it 54th on the November 2017 list, TOP500's own page says 55th. CINES's Adastra press kit dates Occigen as 2015-2022, hence decommissioned in 2022; TOP500 nevertheless kept listing the entry through June 2025, ranking it 443rd in November 2024 (the entry recorded here). Succeeded by adastra.
Timeline
- 2017-01 First operational source
- 2022 Decommissioned source
- 2024-11 List debut at #443 (November 2024) source
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Change history
- 2026-11-01 Added Occigen2 (GENCI/CINES)
Nearby in scale
The machines closest to Occigen2 by Rmax, smaller to larger.
Related systems
Same site
- Occigen · France, upgraded, 2014
- Adastra · France, operational, 2022
- Adastra 2 · France, operational, 2024
- Jade (CINES) · France, upgraded, 2008
Same operator
- Jean Zay V100 · France, operational, 2019
- Turing (IDRIS) · France, decommissioned, 2012
- Jean Zay H100 · France, operational, 2024
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.