On Wed, 2006-03-08 at 08:45 -0800, Stephen A. Smith wrote: > Hi > I'm part of a group that is setting up a Beowulf cluster (64 nodes with > dual procs (3.2GHz Xeons) connected via myrinet) with Scientific Linux. Let us know how you get on with the Myrinet. My own top tip - running gm_allsize without a target machine will do a loopback test. Worth doing if you have a suspect card. Its worth running gm_board_info, or the mx equivalent, on all nodes once they are up.