Computing services > Computing servers
Tehdyt toimenpiteet

Computing servers

CSC has a wide selection of equipment to meet various needs. CSC’s resource allocation group grants academic user quotas for research use based on user account applications.

>> Applying for user accounts

For those who need the most massive computation capacity we have systems that are interconnected via a fast switch network to be used for massive parallel computation.

All essential applications and databases are installed and ready for use on our application server. To supplement the selection we also have dedicated servers for handling of databases and visualization.

Our batch run system controls the computation services dedicated for large- and medium-sized computation. Some of the resources are available for interactive jobs. Detailed instructions are provided in CSC’s user's guides.

Louhi: Cray XT4/XT5

The third phase of the Cray XT supercomputer (louhi.csc.fi) was installed in September 2008. It consists of the earlier XT4 part and the new XT5 part. The XT4 part consists of 1012 and the XT5 part of 1344 quad-core AMD Opteron Barcelona processors. The clock rate of these processors is 2.3 GHz.  In addition, the XT5 part includes the PRACE prototype, Loviatar, upgraded in June 2009, with 360 quad-core AMD Opteron Shanghai processors whose clock rate is 2.7 GHz. The facility totals 10864 compute cores. In general, the amount of memory is 1 GB per core but 384 cores of XT4 and 768 cores of XT5 have 2 GB per core. The total amount of memory is 11.7 TB.

The theoretical peak performance of Louhi is 102.3 TF and the real performance 76.5 TF according to Linpack tests. It is one of the most powerful computers in Europe.

Read more...

Vuori: HP CP4000 BL Proliant supercluster

Vuori (vuori.csc.fi) is a high-performance supercluster. It has 240 compute nodes and each has two six-core 2.6 GHz AMD Opteron processors and thus 2880 cores. The amount of memory in 128 nodes (1536 cores) is 16 GB or 1.33 GB per core, and in 112 nodes (1344 cores)  it is 32 GB or 2.67 GB per core. The theoretical peak performance is 30 Tflop/s.

Read more...

Murska: HP CP4000 BL ProLiant supercluster

Murska (murska.csc.fi) is a high-performance supercluster. It has 544 compute nodes and each node has two dual-core 2.6 GHz AMD Opteron processors. The amount of compute cores is thereby 2*2*544=2176. The theoretical peak performance of Murska is 11.3 Tflop/s.

Read more...

Hippu: HP ProLiant DL580 G7 and DL785 G5 server pair

The application service Hippu (hippu.csc.fi) consists of two newer HP ProLiant DL580 G7 (hippu3.csc.fi and hippu4.csc.fi) and two older DL785 G5 servers (hippu1.csc.fi and hippu2.csc.fi). The newer pair is equipped with four eight-core Intel Xeon X7560 Processors, 2,26 GHz and 1 TB of memory. The older servers provide eight quad-core AMD Opteron 8360 SE (Barcelona) 2.5 GHz processors, and 512 GB of memory each. All together, the service offers 128 cores and 3 TB of memory. The service is aimed for interactive and serial jobs which require a significant amount of memory, and for the usage of the graphical interfaces of the application software. Parallel jobs are recommended to be run on the Vuori super cluster or on the Louhi supercomputer.

Read more...