Sunway TaihuLight node architecture
Sunway TaihuLight is operated by National Supercomputing Center in Wuxi in China. 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
- #27
- HPL Rmax
- 93.01 PFlop/s
- Theoretical peak Rpeak
- 125.44 PFlop/s
- Ranked core count
- 10,649,600
- Measured power
- 15371 kW
- Energy efficiency
- 6.05 GFlop/s per watt
Cluster · Sunway MPP · public profile
- Operator / site
- National Supercomputing Center in Wuxi
- Country
- China
- Profile scope
- Ranked system / partition
- Data checked
- 14 September 2026
- Manufacturer
- NRCPC
- Ranked core count
- 10649600
- Theoretical peak (PFlop/s)
- 125.44
- Measured power (kW)
- 15371
- Energy efficiency (GFlop/s per watt)
- 6.051005139548501
- Operating system
- Sunway RaiseOS 2.0.5
- Network topology
- Sunway Network — three levels: a central switching network between supernodes, a super-node network fully connecting the 256 nodes of each supernode, and a resource-sharing network
Sunway MPP · public profile
- System / node model
- Sunway MPP
- Compute nodes of this type
- 40960
- CPU model
- Sunway SW26010 260C 1.45GHz
- CPU sockets per node
- 1
- Cores per socket
- 260
- Physical CPU cores
- 260
- CPU note
- Each SW26010 is four groups of one management processing element plus 64 computing processing elements — 260 cores in total per node.
- Host / unified memory
- 32 GB DDR3 (4 × 8 GB, 136.51 GB/s)
- Local storage
- None — the report states non-volatile memory is not used in the system
- Form factor
- 256 nodes per supernode · 4 supernodes (1,024 nodes) per cabinet · 40 cabinets
Scope and uncertainty
The operator, NSCC Wuxi, publishes no hardware specification page. Node count, memory, packaging and interconnect follow the Dongarra system report, which is a third-party technical reference rather than an operator statement.
Documented fields: 17 / 22
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
Sunway MPP · public profile — 10 / 15
- ✓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 →
CLUSTER