<div>Dear Pwscf users,</div>
<div> </div>
<div>while running pw.x on a 24 cores clusters (2x2 AMD nodes loaded with RHEL-AS4) some times it is showing the following error after completing few BFGS steps. But the problem is not appearing for all the runs. The error message is</div>
<div> </div>
<div><strong>[cli_0]: aborting job:<br>Fatal error in MPI_Comm_split: Other MPI error, error stack:<br>MPI_Comm_split(250).: MPI_Comm_split(comm=0x84000004, color=0, key=0, new_comm=0x7fbfffdb6c) failed<br>MPIR_Comm_create(79): Too many communicators</strong></div>
<div> </div>
<div>I am using the following command to run pw.x</div>
<div><strong> mpirun -machinefile $HOME/mpd.cores -np 24 pw.x -in <a href="http://file.in">file.in</a> -npool 1 > file.out < /dev/null &</strong></div>
<div> </div>
<div>I have tried by changing the pool number but no use. Please help me in this regard.</div>
<div> </div>
<div>Regards</div>
<div>G. Krishna Chaitanya</div>
<div> </div>