[gmx-users] parallel run hangs (not crashed)

chris.neale at utoronto.ca chris.neale at utoronto.ca
Tue May 16 21:21:48 CEST 2006

>Are you sure you want comm_mode = angular ? It seems like you have a 
>periodic system.

I do have PBC and a quick search indicates that angular is possible with PBC but
can lead to lincs errors or overlaps when using explicit solvent (although I
can't find any reasoning. Anyone know of a reference?) Regardless, I have
switched to linear. Hopefully this is the issue here.

>Are you sure you want to run this system without constraints?

Do you mean all-bonds instead of h-bonds in order to speed things up? My final
system will be much smaller, I am currently using a 4X size with inverted
quadrants 1 and 3 to mimic the P21 asymmetric unit of Dolan Biophys. J. 2002
during membrane equilibration after protein insertion. If you were not
referencing speed, I would benifit from some more explanation.
>If the processes are still there you can kill -HUP them and they should 
>terminate gracefully, writing all files etc.

Thanks for the tip about the -HUP option. And thanks for the lead on comm_mode,
that seems to be a likely suspect.

More information about the gromacs.org_gmx-users mailing list