2018 HPC Upgrade
During the Summer and Fall of 2018, we replaced our previous Penguin cluster with a new cluster from Dell. This upgrade/replacement is part of our normal 3-year cycle.
The current HPC configuration consists of 88 compute nodes with 3520 total CPU cores, 20736 GB total RAM and 61440 total GPU cores:
- Thin: 78 x PowerEdge C6420 Compute Nodes with Intel Skylake Gold 6148 Processors, 192 GB DDR4 RAM and EDR Infiniband.
- GPU: 6 x PowerEdge R740 GPU nodes with Intel Skylake Gold 6148 Processors, 192 GB DDR4 RAM, 2 x NVIDIA V100 GPU and EDR Infiniband - 5120 GPU cores/V100, 10240 GPU cores/GPU node
- Fat: 2 x PowerEdge R740 Fat Memory Nodes with Intel Skylake Gold 6148 Processors, 768 GB DDR4 RAM and EDR Infiniband
- Large Fat: 2 x PowerEdge R740 Fat Memory Nodes with Intel Skylake Gold 6148 Processors,1.5 TB DDR4 RAM and EDR Infiniband
- Parallel File System: DDN GS7K GRIDScalar (GPFS) w/ 120 x 4 TB HDD (480 TB total raw storage) providing up to 8.5GB/sec performance
All compute nodes are connected via EDR Infiniband (2:1 Blocking) and 1GbE for host/OOB management. Head and Login nodes are connected via EDR Infiniband and 10GbE for host/OOB management.