Supercomputers/Systems/Levante GPU Extension
Germany · operational · hpc
Levante GPU Extension
Also known as Levante Grace Hopper partition
Operated by DKRZ at DKRZ (Hamburg) .
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
- 6.75 PFlop/s
- Rpeak
- 8.34 PFlop/s
- Rmax ÷ Rpeak
- 80.9%
- Cores
- 35,904
- Power
- 109.78 kW
- Per watt
- 61.5 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 Levante GPU Extension, and those in Germany. 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 | 6.75 PFlop/s | #142 of 202 · bottom 31% | #20 of 95 · top 20% |
| Energy efficiency | 61.5 GF/W | #16 of 99 · top 15% | #7 of 68 · top 9% |
| Rmax ÷ Rpeak | 80.9 % | #65 of 199 · top 32% | #27 of 94 · top 28% |
| Power draw | 110 kW | #90 of 104 · bottom 15% | #59 of 69 · bottom 16% |
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 | - Eviden (Bull) BullSequana XH3000 | Eviden | - | Verified | dkrz.detop500.org |
| Accelerator | NVIDIA GH200 Grace Hopper Superchip Accelerators Arm CPU + CUDA GPU · TSMC 4N NVIDIA GH200 Superchip (Grace 72C 3GHz) | NVIDIA | - | Verified | dkrz.detop500.org |
| Interconnect | NVIDIA InfiniBand NDR (Quantum-2) Interconnect Fat tree or dragonfly+ · 400 Gb/s per port · Under 0.6 microseconds Quad-Rail NVIDIA InfiniBand NDR200 | NVIDIA | - | Verified | dkrz.detop500.org |
| Operating system | Red Hat Enterprise Linux Operating systems Linux distribution · Open source, commercial support subscription Red Hat Enterprise Linux | Red Hat | - | Reported | top500.org |
The read
A single-rack NVIDIA GH200 Grace Hopper partition added to the German Climate Computing Centre's Levante (see levante, an AMD EPYC 7763 BullSequana XH2000 with an A100 GPU partition), listed by TOP500 as its own system. TOP500 (system 180410) gives BullSequana XH3000, 35,904 cores, quad-rail NVIDIA InfiniBand NDR200, installation year 2025, 6.75 PFlop/s Rmax, 8.34 PFlop/s Rpeak and 109.78 kW (76.83 kW for the optimised run, 5.33 PFlop/s) and ranks it 225th in November 2025. DKRZ's news item of 18 November 2025 calls it a standalone system that is part of Levante, says it placed 3rd on the November 2025 Green500 at 69.426 GFlops/W, tied with two other Eviden XH3000 systems, and quotes 227th on the TOP500 where the list page shows 225th. Neither source states the number of superchips or nodes or a go-live date, and DKRZ's HPC systems page still describes only the A100 GPU partition, so the confidence stays reported. Used for energy-efficient climate simulation at the Max Planck Institute for Meteorology per DKRZ. DKRZ's planned successor HLRE-5 is not covered here.
Timeline
Public list appearances
Pointers only: rank and edition, linking to the canonical entry. We do not reproduce list tables. See sourcing policy.
Site & facility
Change history
- 2026-10-29 Added Levante GPU Extension
Nearby in scale
The machines closest to Levante GPU Extension by Rmax, smaller to larger.
- Camphor 3 · Japan, operational · 6.71 PFlop/s
- Christofari · Russia, operational · 6.67 PFlop/s
- ECMWF HPCF-AA · Italy, operational · 6.67 PFlop/s
- Karolina · Czechia, operational · 6.75 PFlop/s
- Vega · Slovenia, operational · 6.92 PFlop/s
- Gaia · Brazil, operational · 6.97 PFlop/s
Related systems
Same site
- Levante · Germany, operational, 2021
- DKRZ Mistral · Germany, operational, 2015
Same accelerator
- DAEDALUS · Greece, under construction
- Discovery 6 · USA, operational, 2025
- Isambard-AI Phase 2 · United Kingdom, operational, 2025
- JUPITER · Germany, operational, 2025
- KAIROS · France, operational
- KISTI-6 · South Korea, under construction
Further reading
Operating a supercomputer
Acceptance testing, why "installed" and "in production" are six months apart, and the machine lifecycle.
Inside a node
Why accelerators exist, what the memory hierarchy costs, and how the CPU and accelerator merged onto one package.
The interconnect
Topologies, why latency and tail behaviour matter more than bandwidth, and how the fabric market consolidated.