New deploy: two jobs per node

by **Admin** Fri Sep 19, 2014 4:29 pm

With our deployment of this afternoon, we are finally running two jobs per node (one on each of the two quad-core processors on our compute nodes). This should effectively double the throughput of StarExec, since our previous version of the system just left one processor, and half the memory, idle on each node at all times. Now we will utilize all our hardware. Please let us know if you see anomalous behavior with this change. We have tested this thoroughly on our development cluster, and it seems to be working fine currently on StarExec.
Aaron