Gefion

Danish Centre for AI Innovation

Node type

CPU Accelerator Interconnect

Reading this map

Gefion node architecture

Gefion is operated by Danish Centre for AI Innovation in Denmark. This profile covers Ranked system / partition. Specifications are listed below; anything the operator has not published is itemised in the documented-fields list.

Published benchmark (TOP500 · June 2026)

HPL rank: #43
HPL Rmax: 66.59 PFlop/s
Theoretical peak Rpeak: 100.63 PFlop/s
Ranked core count: 223,088
Measured power: 1753 kW
Energy efficiency: 37.99 GFlop/s per watt

Cluster · DGX H100 platform

Operator / site: Danish Centre for AI Innovation
Country: Denmark
Profile scope: Ranked system / partition
Data checked: 14 September 2026
Manufacturer: Nvidia
Ranked core count: 223088
Theoretical peak (PFlop/s): 100.63
Measured power (kW): 1753
Energy efficiency (GFlop/s per watt): 37.98630918425556
Operating system: Red Hat Enterprise Linux
Network topology: Octo-rail InfiniBand (eight-rail, rail-optimised)

DGX H100 platform

Items marked ▧ come from the documented NVIDIA DGX H100 platform specification.

System / node model: NVIDIA DGX H100
Compute nodes of this type: 191
CPU model: Xeon Platinum 8480C 56C 2GHz
CPU sockets per node: 2
Cores per socket: 56
Physical CPU cores: 112
Accelerator model: NVIDIA H100 SXM5 80GB
Physical accelerators per node: 8
Host / unified memory: 2 TB (DGX H100 platform specification)
Accelerator memory: 80 GB per GPU
Accelerator ↔ accelerator link: NVLink / NVSwitch
On-node fabric: NVLink 4 through 4 × 4th-generation NVSwitch, 900 GB/s GPU-to-GPU
CPU ↔ accelerator link: PCIe 5.0
Network interfaces per node: 8
Interface model: NVIDIA ConnectX-7 single-port InfiniBand
Interface speed: 400 Gb/s InfiniBand per adapter
Interface-to-accelerator topology: One adapter per GPU
Management / storage networking: 2 × ConnectX-7 dual-port Ethernet (storage and in-band management) + 1 GbE BMC
Local storage: 2 × 1.92 TB NVMe M.2 SSD in RAID 1 (OS) + 8 × 3.84 TB NVMe U.2 SSD in RAID 0 (data cache)
Form factor: 8U rackmount · 6 × 3.3 kW power supplies (4+2 redundant)

Scope and uncertainty

Node count follows the DCAI announcement: 191 NVIDIA DGX H100 units totalling 1,528 H100 GPUs. The fabric is recorded as the rail structure TOP500 states; note that TOP500 names the interconnect HDR100 (100 Gb/s) while DCAI and NVIDIA describe Quantum-2 InfiniBand (NDR, 400 Gb/s), so the generation is unresolved. Node specifications follow the documented NVIDIA DGX H100 platform (8U rackmount, 6 × 3.3 kW power supplies (4+2 redundant)); operator-specific deviations are not published.

Documented fields: 27 / 29

Every field this catalog checks. Fields listed above have a value; the remainder are shown here. A field marked absent is not publicly available in the sources we could verify, which does not mean the hardware is missing. Fields that cannot apply, such as benchmark results for an unranked system, are excluded from the total.

Cluster — 7 / 7

Manufacturer
Ranked core count
Theoretical peak (PFlop/s)
Measured power (kW)
Energy efficiency (GFlop/s per watt)
Operating system
Network topology

DGX H100 platform — 20 / 22

System / node model
Compute nodes of this type
CPU model
CPU sockets per node
Cores per socket
Physical CPU cores
CPU note
Accelerator model
Physical accelerators per node
Host / unified memory
Accelerator memory
Accelerator ↔ accelerator link
On-node fabric
CPU ↔ accelerator link
Network interfaces per node
Interface model
Interface speed
Interface-to-accelerator topology
Aggregate scale-out per node
GPUDirect RDMA
Management / storage networking
Local storage
NVMe data disks
Max remote data disks
Uncached disk IOPS
Uncached disk throughput (MBps)
Form factor

Sources

Browse all cluster architecture pages →