The “Tai-Yi” HPC Cluster
Overall Peak Performance: 2.5 PFlops
Measured Continuable Performance: 1.687 PFlops
Login Nodes: 4
Spec: 2 Intel Xeon Gold 6140 CPU (2.3GHz/18c), 384 GB RAM
Blade Nodes: 815
Spec: 2 Intel Xeon Gold 6148 CPU (2.4GHz/18c), 192 GB RAM
Flat Nodes: 2
Spec: 8 Intel Xeon Platinum 8160 CPU (2.1GHz/24c), 6 TB RAM
GPU Nodes: 4
Spec: 2 Intel Xeon Gold 6148 CPU (2.4GHz/20c), 384 GB RAM, 2 NVIDIA V100 GPU
Storage: 5.5 PB
Spec: GPFS Parallel File System, Measured I/O throughput exceeds 40 GB/s
The “Qi-Ming” HPC Cluster
Overall Peak Performance: 0.3 PFlops
Measured Continuable Performance: 0.191 PFlops
Login Nodes: 4
Spec: 2 Intel Xeon E5-2690v3 CPU (2.6GHz/12c), 128 GB RAM
Blade Nodes: 230
Spec: 2 Intel Xeon E5-2690v3 CPU (2.6GHz/12c), 64 GB RAM
Flat Nodes: 7
Spec: 8 Intel Xeon E7-8880v3 CPU (2.6GHz/12c), 6 TB RAM
GPU Nodes: 6
Spec: 2 Intel Xeon E5-2690v3 CPU (2.6GHz/12c), 128 GB RAM, 4 NVIDIA Tesla K80 GPU
Storage: 528TB
Spec: Lustre Parallel File System, Measured I/O throughput exceeds 6 GB/s