By Costin Caramarcu |
Tue, 01/26/2021 - 20:04
About the Institutional Cluster (IC) at the SDCC.
Prerequisites:
- Have a valid account with the SDCC
- Apply here
- Have a valid account in slurm
- Your liaison should contact us with your name or user id via the ticketing system
Cluster Information:
The cluster consists of:
- 216 worker nodes
- 2 submit nodes
- 2 master nodes
The worker nodes detail:
- HPE ProLiant XL190r Gen9
- 2 CPUs Intel(R) Xeon(R) CPU E5-2695 v4 @ 2.10GHz
- NUMA node0 CPU(s): 0-8,18-26
- NUMA node1 CPU(s): 9-17,27-35
- Thread(s) per core: 1
- Core(s) per socket: 18
- Socket(s): 2
- NUMA node(s): 2
- 2x Nvidia K80 (108 worker nodes) or 2x P100 (108 worker nodes) per node (4 K80 or 2 P100 devices per node)
- 256 GB Memory
- InfiniBand EDR connectivity
Storage:
- 1.9 TB of local disk storage per node
- 1 PB of GPFS distributed storage
- Cluster Storage
Partitions:
| partition | time limit | allowed qos | default time | preempt mode | user availability |
|---|---|---|---|---|---|
| debug | 30 minutes | normal | 5 minutes | off | ic |
| long | 24 hours | normal | 5 minutes | off | ic |
| scavenger | 6 hours | scavenger | 5 minutes | cancel | ic |
| sky | 24 hours | sky | 5 minutes | off | lqcd only |
| gen3 | 72 hours | gen3 | 2 hours | off | cfn only |
| gen4 | 72 hours | gen4 | 2 hours | off | cfn only |
Debug partition:
The debug partition consists of 2 x HPE ProLiant XL190r Gen9 with Nvidia K80
Limits:
- Each user can submit a maximum of 50 jobs
- The maximum number of nodes running jobs for an account varies with the size of the allocation.
Licensed Software
The institutional cluster offers only 2 licensed software:
- Intel Parallel Studio XE
- PGI Compiler
Any other licensed software that you may see available via modules is installed and maintained by various groups and SDCC/CSI does not offer support for it.
