Shaheen III - GPU node architecture
Shaheen III - GPU is operated by King Abdullah University of Science and Technology in Saudi Arabia. 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
- #22
- HPL Rmax
- 122.8 PFlop/s
- Theoretical peak Rpeak
- 155.21 PFlop/s
- Ranked core count
- 574,464
- Measured power
- 1980 kW
- Energy efficiency
- 62.02 GFlop/s per watt
Cluster · GPU - HPE Cray EX254n · public profile
- Operator / site
- King Abdullah University of Science and Technology
- Country
- Saudi Arabia
- Profile scope
- Ranked system / partition
- Data checked
- 14 September 2026
- Manufacturer
- HPE
- Ranked core count
- 574464
- Theoretical peak (PFlop/s)
- 155.21
- Measured power (kW)
- 1980
- Energy efficiency (GFlop/s per watt)
- 62.02020202020202
- Operating system
- HPE Cray OS
- Network topology
- Two-level Dragonfly
GPU - HPE Cray EX254n · public profile
Items marked ▧ come from the documented NVIDIA GH200 Grace Hopper Superchip platform specification.
- System / node model
- GPU - HPE Cray EX254n
- Compute nodes of this type
- 704
- CPU model
- NVIDIA Grace 72C 3.1GHz
- CPU sockets per node
- 4
- Cores per socket
- 72
- Physical CPU cores
- 288
- Accelerator model
- NVIDIA GH200 Superchip
- Physical accelerators per node
- 4
- CPU ↔ accelerator link
- NVLink-C2C · 900 GB/s coherent CPU ↔ GPU
- Form factor
- 7 HPE Cray EX4000 cabinets · 16 Slingshot switches per cabinet · 64-port switch radix
Scope and uncertainty
KAUST states 704 GPU compute nodes with four Grace Hopper Superchips each, which implies 2,816 superchips; its own documentation page instead states 2,800. The two cannot both be right, and KAUST does not reconcile them. 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: 17 / 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
GPU - HPE Cray EX254n · public profile — 10 / 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 →
CLUSTERATLAS