High-Performance-Computing-Cluster dirac
The high performance computing cluster dirac (named after Nobel laureate in physics Paul Dirac) was delivered in the end of 2006 and is operational since 18.01.2007. The cluster was extended six times until now.
It has one access- and control node (master node) dirac and 29 compute nodes with 692 CPU cores and 1022 GiB RAM. The hardware and user environment is optimized for parallel calculations. Two of the compute nodes, named dinux4 and dinux5, are in operation as universal Linux servers, another one has been reserved for special purposes. The remaining 26 compute nodes are operated under the control of the batch system Sun Grid Engine.
The software environment working above the operating systems openSUSE Linux 10.1 (64 bit) and 11.3 (on the nine Magny Cours servers) as operating systems contains optimizing compilers, mathematical libraries, runtime environments for parallel development and program execution and some popular applications for physics and chemistry.
Every user with a HZB account (internal password) is eligible to use the cluster without any further action. They will need a ssh connection with ssh (Linux/Unix/MacOS) or PuTTY (Windows) to the master node dirac only. In addition to the general rules for IT in HZB, some special rules have been released.
As a recommendation, potential users should contact FM-D first. For answers about the cluster and its operation please read the hints for usage or contact Andreas Tomiak, issues about software matters may be answered by Michael Fromme.
(Kopie 1)
[at]
