HOME > Systems > SR16000 > Introduction of SR16000 System

Introduction of SR16000 System

The SR16000 model M1 (Yayoi) consists of Power7 computational nodes. Each node has four Power7 processors and 200GB of shared memory. Each of eight nodes are connected to each other via a fast network.
We began providing computing service (only for the “personal course”) in October, 2011 as the successor system of the SR11000.
This system is expected to achieve research outcomes for many existing programs which require large shared memory.

Hardware

Theoretical Computing Speed 54.906 TFlops
Number of nodes 56 nodes
Memory Capacity 11,200 GB
Theoretical Computing Speed per node 980.48 GFlops
Memory Capacity per node 200 GB
Disk capacity 556 TB
Network Hierarchical Fullbisection

System conceptual diagram

Software

OS IBM AIX 7.1
Compiler HITACHI (Fortran 77/90, C, C++)
IBM (XL C/C++, JAVA)
GNU, g95
Batch System HITACHI NQS
Filesystem GPFS (IBM General Parallel File System for AIX)
Libraries MATRIX/MPP, MATRIX/MPP/SSS
MSL2, BLAS, LAPACK, ScaLAPACK, FFTW,
Parallel NetCDF,SuperLU, SuperLU_DIST,
ESSL, Parallel ESSL
Applications Gaussian09