Hi CCPEM
Regarding the cluster node configuration described in http://www.cryoem.se/cluster-nodes-for-relion-gpu/ . Is it correct to assume that the users would be connecting to the node via an X over ssh ? in which case the minimizing network latency would be critical ?( if so what kind speed of Infiniband are you using?). Is this system connected to a storage server to compensate for the storage drawback mentioned ?
Any inputs regarding the network and storage specifics would really help me in deciding to the rack server form factor over a GPU workstation
Sincerely
Joshua Lobo
|