Bi
The Bi cluster is operated for SMHI research and development. It follows in the footsteps of the earlier clusters Gimle, Dunder, and Krypton.
The Bi cluster has a LINPACK performance of approximately 326 Teraflop/sec using all 10,256 compute cores.
The system was delivered in the end of 2014 by ClusterVision and made available to users in March 2015.
User documentation
System Details
| Hardware | Custom design based on ASUS servers |
| Processors | 8-core Intel Xeon E5-2640v3 processors at 2.6 GHz |
| Number of compute nodes | 641 |
| Compute nodes (thin) | 2 sockets (16 cores) and 64 GB memory |
| Compute nodes (fat) | 2 sockets (16 cores) and 256 GB memory |
| Login node | 2 sockets (16 cores) and 256 GB memory |
| High speed interconnect | Intel Truescale Infiniband QDR |
| Node scratch storage | 270 TiB (420 GB per node) |
| Global file system | SMHI's NFS and Lustre file systems /home and /nobackup |
| Operating system | CentOS Linux 7 (upgraded from CentOS 6 during 2020) |
| Batch queue system | Slurm |
Research performed with Bi
Bi is used by the research division of SMHI, primarily for applications in climate, meteorology and oceanography. In the climate area, SMHI collaborates with European partners in the development, analysis and use of both global and regional climate models, and with impact studies in the fields of oceanography, hydrology and air quality.
Who can use Bi?
Access to Bi is granted by SMHI.
User Area
