There seems not to be any problem in your arch make if openmpi is used and compiled with the intel compiler.
Else, you need to do debugging on the specific machine you have problems with, consult with your cluster-administrator to best debug this. But it could be related to some kind of memory leak (check that). And have you set ulimit? Which openmpi version are you using? In my installation 1.8.7 had a memory leak, so refrain from using that version. 2015-10-31 15:09 GMT+01:00 Xiaoming Wang <[email protected]>: > Dear siesta users and developers, > > > > I have a problem when the scf loop seems going to be finished, the code > crashed. See the following. However the same input works on another > computer. So does anyone know what’s going on here? I attached my > arch.make. Any comment is appreciated. > > > > *************************************************************************** > > siesta: 106 -152559.7801 -152559.7799 -152559.8859 0.0000 -2.4612 > > siesta: 107 -152559.7801 -152559.7799 -152559.8859 0.0000 -2.4612 > > siesta: 108 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > siesta: 109 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > siesta: 110 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > siesta: 111 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > siesta: 112 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > siesta: 113 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > siesta: 114 -152559.7801 -152559.7800 -152559.8860 0.0000 -2.4612 > > -------------------------------------------------------------------------- > > orterun noticed that process rank 3 with PID 24733 on node comet-23-62 > exited on signal 11 (Segmentation fault). > > -------------------------------------------------------------------------- > > -------------------------------------------------------------------------- > > WARNING: A process refused to die despite all the efforts! > > This process may still be running and/or consuming resources. > > > > Host: comet-23-62 > > PID: 24759 > > > > -------------------------------------------------------------------------- > > IBRUN: Job ended with value 139 > > > > > > Xiaoming Wang > > Rutgers University > -- Kind regards Nick
