MPI parallelisation fails when using over a certain number of cores

[Noah Harrison]

Hi,

I'm encountering a weird issue when using gmx_MMPBSA.

When I run the following command, I encounter no errors and calculations complete normally:

• gmx_MMPBSA -O -i ../../mmbpsa.in -cs ../nvt.tpr -ci ../index.ndx -cg 1 13 -ct ../nvt.xtc -lm ../Ligand.mol2 -nogui

When I use mpi to parallelise with 2 or 4 cores, the calculations run correctly. However, when I run using 8 cores, the calculation fails. I have attached the output displayed when I run the following command:

•  mpirun -np 8 gmx_MMPBSA -O -i ../../mmbpsa.in -cs ../nvt.tpr -ci ../index.ndx -cg 1 13 -ct ../nvt.xtc -lm ../Ligand.mol2 -nogu

Any ideas what might be causing this issue? Am I using mpirun incorrectly?

Thanks,

Noah

[marioe911116]

Add MPI to the command line after gmx_MMPBSA and it should run smoothly:

mpirun -np 8 gmx_MMPBSA MPI -O -i ../../mmbpsa.in -cs ../nvt.tpr -ci ../index.ndx -cg 1 13 -ct ../nvt.xtc -lm ../Ligand.mol2 -nogui

cheers!