Arina has 3.728\u00a0cores<\/strong> distributed in 3.664\u00a0xeon cores, and 64\u00a0Itanium2 cores with RAM memory ranging from 16 to 512 GB per node. Four of the xeon nodes have 2 Nvidia Tesla cards each, another node with \u00a0two Nvidia Kepler 20, and another one with 2 Nvidia Kepler 40 . There are 3 high performance file systems based on Lustre, one with 4.2 TB for the Itanium2 and Opteron nodes and other 2 with 22 and 40 TB for the Xeon nodes. All the nodes are connected with a high bandwidth and low latency Infiniband network.<\/p>\n Arina has 3 servers, arina<\/strong> (Itanium2), katramila<\/strong> and guinness<\/strong> (Xeon) where researches can use to access the cluster. This servers work as connection, visualization, compilation and test servers. All the servers mount the same They are similar to the previous ones:<\/p>\n Specific features of the nodes:<\/p>\n The labels are used to select specific computing nodes.<\/p>\n Specific features of the nodes:<\/p>\n All the Itanium2 computing nodes have the itanium<\/em> label. The labels are used to select specific computing nodes.<\/p>\n In addition to the local file systems (\/scratch<\/tt>) nodes there are three shared file systems.<\/p>\n General information Summary Arina has 3.728\u00a0cores distributed in 3.664\u00a0xeon cores, and 64\u00a0Itanium2 cores with RAM memory ranging from 16 to 512 GB per node. Four of the xeon nodes have 2 Nvidia Tesla cards each, another node with \u00a0two Nvidia Kepler 20, and another one with 2 Nvidia Kepler 40 . There are 3 high … Seguir leyendo Arina Cluster<\/span> Other resources<\/h3>\n
\/home<\/code> file system using NFS. Every night the\/home is backed up. <\/code> When a computing node writes in \/home<\/code> the data is sent using the NFS protocol using an ethernet \u00a0slow connection. We strongly recommend to use during the calculations the \u00a0the local \/scratch<\/code> or the shared \/gscratch<\/code> filesystems to storage temporary data.<\/p>\n<\/code><\/p>\nFeatures of the Xeon nodes<\/h2>\n
\n
Nodes with GPGPUs<\/h3>\n
\n
\n\n
\n Amount<\/td>\n Name<\/td>\n Type<\/td>\n Proc<\/td>\n Cores<\/td>\n Speed<\/td>\n RAM (GB)<\/td>\n Disk (GB)<\/td>\n Label<\/td>\n<\/tr>\n \n 1<\/td>\n Guinness<\/td>\n Server<\/td>\n 2<\/td>\n 8<\/td>\n 2.3 GHz<\/td>\n 12<\/td>\n —<\/td>\n —<\/td>\n<\/tr>\n \n 1<\/td>\n Katramila<\/td>\n Server<\/td>\n 2<\/td>\n 8<\/td>\n 2.6 GHz<\/td>\n 128<\/td>\n —<\/td>\n —<\/td>\n<\/tr>\n \n 1<\/td>\n cn3<\/td>\n Node<\/td>\n 2<\/td>\n 8<\/td>\n 2.3 GHz<\/td>\n 96<\/td>\n 250<\/td>\n xeon,xeon8<\/td>\n<\/tr>\n \n 17<\/td>\n cn4-20<\/td>\n Node<\/td>\n 2<\/td>\n 8<\/td>\n 2.3 GHz<\/td>\n 24<\/td>\n 250<\/td>\n xeon,xeon8<\/td>\n<\/tr>\n \n 36<\/td>\n cn21-56<\/td>\n Node<\/td>\n 2<\/td>\n 8<\/td>\n 2.3 GHz<\/td>\n 48<\/td>\n 250<\/td>\n xeon,xeon8<\/td>\n<\/tr>\n \n 4<\/td>\n cn57-60<\/td>\n GPU node<\/td>\n 2<\/td>\n 8<\/td>\n 2.3 GHz<\/td>\n 24<\/td>\n 250<\/td>\n xeon,xeon8,gpu<\/td>\n<\/tr>\n \n 18<\/td>\n cn61-78<\/td>\n Node<\/td>\n 2<\/td>\n 12<\/td>\n 2.4 GHz<\/td>\n 48<\/td>\n 250<\/td>\n xeon,xeon12<\/td>\n<\/tr>\n \n 1<\/td>\n cn79<\/td>\n Node<\/td>\n 2<\/td>\n 12<\/td>\n 2.4 GHz<\/td>\n 96<\/td>\n 250<\/td>\n xeon,xeon12<\/td>\n<\/tr>\n \n 35<\/td>\n cn80-cn114<\/td>\n Node<\/td>\n 2<\/td>\n 12<\/td>\n 2.4 GHz<\/td>\n 24<\/td>\n 250<\/td>\n xeon,xeon12<\/td>\n<\/tr>\n \n 30<\/td>\n nb1-nb30<\/td>\n Node<\/td>\n 2<\/td>\n 20<\/td>\n 2.8 GHz<\/td>\n 128<\/td>\n 128<\/td>\n xeon,xeon20<\/td>\n<\/tr>\n \n 1<\/td>\n nb31<\/td>\n Node<\/td>\n 2<\/td>\n 32<\/td>\n 2.6 GHz<\/td>\n 512<\/td>\n 1000<\/td>\n xeon,xeon20<\/td>\n<\/tr>\n \n 1<\/td>\n nb32<\/td>\n GPU node<\/td>\n 2<\/td>\n 20<\/td>\n 2.8 GHz<\/td>\n 128<\/td>\n 128<\/td>\n xeon,xeon20,gpu,rh7<\/td>\n<\/tr>\n \n 66<\/td>\n nd1-nd14,nd16-nd67<\/td>\n Node<\/td>\n 2<\/td>\n 28<\/td>\n 2.4 GHz<\/td>\n 128<\/td>\n 128<\/td>\n xeon,xeon28,rh7<\/td>\n<\/tr>\n \n 1<\/td>\n nd15<\/td>\n GPU Node<\/td>\n 2<\/td>\n 28<\/td>\n 2.4 GHz<\/td>\n 128<\/td>\n 128<\/td>\n xeon,xeon28,rh7,gpu<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n Features of the Itanium2 nodes<\/h2>\n
\n
\n\n
\n Amount<\/td>\n Name<\/td>\n Type<\/td>\n Proc<\/td>\n Cores<\/td>\n Speed<\/td>\n RAM (GB)<\/td>\n Disk (GB)<\/td>\n Label<\/td>\n<\/tr>\n \n 1<\/td>\n Arina<\/td>\n Server<\/td>\n 2<\/td>\n 4<\/td>\n 1.6 GHz<\/td>\n 8<\/td>\n —<\/td>\n —<\/td>\n<\/tr>\n \n 1<\/td>\n Arinaz<\/td>\n Server<\/td>\n 2<\/td>\n 2<\/td>\n 1.3 GHz<\/td>\n 2<\/td>\n —<\/td>\n —<\/td>\n<\/tr>\n \n 4<\/td>\n cndXX<\/td>\n Node<\/td>\n 4<\/td>\n 8<\/td>\n 1.6 GHz<\/td>\n 16<\/td>\n 550<\/td>\n itaniumb<\/td>\n<\/tr>\n \n 10<\/td>\n cndXX<\/td>\n Node<\/td>\n 4<\/td>\n 8<\/td>\n 1.6 GHz<\/td>\n 32<\/td>\n 550<\/td>\n itaniumb<\/td>\n<\/tr>\n \n 1<\/td>\n cnd43<\/td>\n Node<\/td>\n 4<\/td>\n 8<\/td>\n 1.6 GHz<\/td>\n 64<\/td>\n 550<\/td>\n itaniumb<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n <\/h2>\n
High performance filesystems<\/h2>\n
\n