The Albert100 and Albert2 systems are the result of a Parma-Rome-Trieste collaboration and were for a long time our computational power horses. They were dismised at the beging of 2010 and for historical reasones will still provide information on the two systems.

The Albert2 cluster

Parallel cluster (Albert2) with 16 computing nodes and 3 TBytes of RAID 5 disk storage. Each computing node is a bi-processor Opteron 2 GHz with 4 GBytes of RAM and Infiniband high-speed interconnections. The system was co-financed by the MIUR in 2004 and it is on-line ever since.

The Albert 100 clusters

The Albert 100 system is the result of a Parma-Rome-Trieste collaboration and was financed by a special INFN "Commisione IV" grant in 2001. The Albert100 system was originally made up of a master node connected to 44 computing nodes (now 32) using 100BaseT fast ethernet communications. Although its performances are not state-of-the-art is still a very useful workhorse for code-developing and testing and to perform small-computations.

Albert 100 has a theoretical peak performance of 100 GFlops and it is dedicated to the numerical simulations of Einstein's General Relativity field equations. The performance of the system were optimized for the use of the Cactus/Whisky numerical framework and the results of the system benchmarks are shown here.

In March 2008 we start the migration of the Albert100 cluster to the INFN grid. This is part of the plan to migrate all the numerical simulations of the group to the GRID infrastructure.