High Performance Intercloud - Hokkaido University Information Initiative Center

HOKKAIDO UNIVERSITY
SUPERCOMPUTER
HOME > SUPERCOMPUTER >

Overview of Supercomputer system

Overview of Supercomputer system

System configuration

Supercomputer system consists of Grand Chariot (Subsystem A), Polaire (Subsystem B), and the Supercomputer storage system.

supercomputer system

System specifications
Grand Chariot (Subsystem A)
System Machine name PRIMERGY CX2550 M4,PRIMERGY CX400 M4
Number of nodes 1,004
Peak FLOPS
(Double precision)
3.08PFLOPS
Total memory amount 376.5TB
Network topology Full-bisection Fat Tree
Node Number of CPUs
(Number of cores)
2 (2×20=40)
Peak FLOPS
(Double precision)
3.072TFLOPS
(2.4GHz×20-core×32-operation×2CPU)
Total memory amount 384GB
Peak memory bandwidth 255.9GB/s
(2,666MHz×8B×6ch×2CPU)
Storage in node 240GB SSD × 1
Interconnect Omni-Path × 2
(12.5GB/s × 2)
OS CentOS
CPU CPU name Intel Xeon Gold 6148
Architecture Skylake
Clock 2.4GHz
(Up to 3.7GHz by Turbo boost )
Number of cores 20
Peak FLOPS
(Double precision)
1.536TFLOPS
Polaire (Subsystem B)
System Machine name PRIMERGY CX1640 M1, PRIMERGY CX600 M1
Number of nodes 288
Peak FLOPS
(Double precision)
0.87PFLOPS
Total memory amount 31.5TB
Network topology Full-bisection Fat Tree
Node Number of CPUs
(Number of cores)
1 (1×68=68)
Peak FLOPS
(Double precision)
3.0464TFLOPS
(1.4GHz×68-core×32-operation×1CPU)
Total memory amount 96GB
Peak memory bandwidth 115.2GB/s
(2,400MHz×8B×6ch×1CPU)
Total memory amount
(High-speed memory)
16GB MCDRAM
Peak memory bandwidth
(High-speed memory)
921.6GB/s
(7.2GT/s×8B×8chip×2(R/W))
Storage in node 64GB SATA Flush × 1
Interconnect Omni-Path × 1
(12.5GB/s × 1)
OS CentOS
CPU CPU name Intel Xeon Phi 7250
Architecture Knights Landing
Clock 1.4GHz
(Up to 1.6GHz by Turbo boost)
Number of cores 68
Peak FLOPS
(Double precision)
3.0464TFLOPS
Supercomputer storage system
Machine name DDN ES14KX
File system DDN EXAScaler (Lustre)
Total capacity 16PB
Data transfer speed 160GB/s