[Pw_forum] (no subject)
akohlmey at cmm.chem.upenn.edu
Mon Mar 24 03:07:06 CET 2008
On Mon, 24 Mar 2008, wrote:
YC> Dear all:
YC> I've successfully compiled parallel version pwscf 3.2.3 &4.0 on my
YC> system (suse10 for IA64). The included tests are passed. But I can't run
YC> pw.x with mpi.
YC> To clearly show my problem, I describe it as follows:
YC> first, I test the mpi to check whether it can work. It works!
how did you test it?
YC> second, I test the pwscf. I run pw.x:$pw.x < al.scf.in >al.scf.out&. it can
are you _sure_ you compiled a parallel version?
most MPI libraries do not allow the parallel executable
to be run without the MPI environment, ie. via mpirun.
YC> third, I test the mpi with pwscf. I run:$ mpirun -np 10 pw.x < al.scf.in >
YC> al.scf. out &. it can't wrok.
YC> The errors are:Missing: program name Program pw.x either does not
YC> exist, is not executable, or is an erroneous argument to mpirun.
is pw.x in your regular search path for exectuables?
try: which pw.x
also you can try specifying the full path to pw.x
finally, you should not use I/O redirection with parallel
execution. this is not handled portably across MPI implementations.
better to use the -in flag. to specify the input filename.
YC> Is there any suggestions to solve this problem?
YC> sincerely yours
YC> Cheng YC
Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu http://www.cmm.upenn.edu
Center for Molecular Modeling -- University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582, fax: 1-215-573-6233, office-tel: 1-215-898-5425
If you make something idiot-proof, the universe creates a better idiot.
More information about the users