CLUSTERATLAS
A field guide to supercomputing

Arrhenius GPU

EuroHPC/NAISS (National Academic Infrastructure for Super­computing in Sweden)

CPU Accelerator Interconnect
i
Reading this map

Arrhenius GPU node architecture

Arrhenius GPU is operated by EuroHPC/NAISS (National Academic Infrastructure for Super­computing in Sweden) in Sweden. 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
#42
HPL Rmax
66.82 PFlop/s
Theoretical peak Rpeak
84.02 PFlop/s
Ranked core count
310,080
Measured power
1088 kW
Energy efficiency
61.42 GFlop/s per watt

Cluster · GPU compute

Operator / site
EuroHPC/NAISS (National Academic Infrastructure for Super­computing in Sweden)
Country
Sweden
Profile scope
Ranked system / partition
Data checked
14 September 2026
Manufacturer
HPE
Ranked core count
310080
Theoretical peak (PFlop/s)
84.02
Measured power (kW)
1088
Energy efficiency (GFlop/s per watt)
61.41544117647059
Operating system
HPE Cray OS

GPU compute

Items marked ▧ come from the documented NVIDIA GH200 Grace Hopper Superchip platform specification.

System / node model
HPE Cray EX254n
Compute nodes of this type
382
CPU model
NVIDIA Grace · 72 Arm Neoverse V2 cores per superchip
CPU sockets per node
4
Accelerator model
NVIDIA GH200 Superchip
Physical accelerators per node
4
Host / unified memory
128 GB LPDDR5X per module (4 modules per node)
Accelerator memory
96 GB HBM3 per GPU
CPU ↔ accelerator link
NVLink-C2C · 900 GB/s coherent CPU ↔ GPU
Network interfaces per node
4
Interface model
HPE Slingshot
Interface speed
200 Gb/s per adapter
Local storage
1.8 TB NVMe node-local scratch

Scope and uncertainty

NAISS lists the interconnect as "4x Slingshot 200 GB"; the unit as written is ambiguous, so a per-adapter rate is not recorded. NAISS also gives 29 PB for the parallel file system on its resource page against 25 PB disk plus 2 PB flash in the technical description, and does not reconcile the two. Node specifications follow the documented NVIDIA GH200 Grace Hopper Superchip platform (Superchip module — nodes per system and per-node superchip counts are operator-specific); operator-specific deviations are not published.

Documented fields: 19 / 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 — 6 / 7

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

GPU compute — 13 / 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 →

Data & sources