Can you send me the compile environment. I.e., what environment variables have you set for compiling NWChem? Seems like you build NWChem in serial.
Thanks,
Bert
Quote:Lumbrius Apr 19th 1:15 pmHello dear NWChem developers and users.
I cant make NWChem work correctly with MPI.
I have Open Mpi version 1.4.3
when I attempt to run next same test file
start
title "WATER 6-311G* meta-GGA XC geometry"
echo
geometry units angstroms
O 0.0 0.0 0.0
H 0.0 0.0 1.0
H 0.0 1.0 0.0
end
basis
H library 3-21g
O library 3-21g
end
dft
iterations 100
print kinetic_energy
xc xtpss03 ctpss03
decomp
end
task dft optimize
with 2 different commands, first of them is
mpirun -np 1 nwchem water1.inp >& water1.out
i.e., serial execution
second is
mpirun -np 2 nwchem water2.inp >& water-2.out &
I get very wrong result in last case:
lucius@diana ~/nwchem-6.0/input $ ls -la *out
-rw-r--r-- 1 lucius users 94171 ??? 19 17:13 water1.out
-rw-r--r-- 1 lucius users 222714 ??? 19 17:14 water2.out
seems two independent processes are executed and they just do double work and output into log.
Could you tell me please how to make NWChem to execute in parallel correctly? |