Il 29/01/21 15:58, Gilles Gouaillardet via users ha scritto: Hi Gilles.
Tks for the answer. > the mpirun command line starts 2 MPI task, but the error log mentions > rank 56, so unless there is a copy/paste error, this is highly > suspicious. Uhm... Going to re-check. Most probably it's just my error substituting a variable, but worth checking again. > I invite you to check the filesystem usage on this node, and make sure > there is a similar amount of available space in /tmp and /dev/shm (or > other filesystem if you use a non standard $TMPDIR Well, on all those nodes /tmp is disk-based (~52G available) while /dev/shm is a tmpfs w/ 239G "available" mounted as tmpfs /dev/shm tmpfs defaults,size=95% (we've had to increase the default to 95% because that's required by the new release of a library that "mixes" MPI and OpenMP to squeeze a bit more speed reducing comms overhead). -- Diego Zuccato DIFA - Dip. di Fisica e Astronomia Servizi Informatici Alma Mater Studiorum - Università di Bologna V.le Berti-Pichat 6/2 - 40127 Bologna - Italy tel.: +39 051 20 95786
