[Pw_forum] use doual core

[Axel Kohlmeyer]

On Sun, 10 Dec 2006, Amin Babazadeh wrote:

dear amin,

[...]

AB> # See include/defs.h.README for a list of options and their meaning
AB> # With the exception of IBM xlf, FDFLAGS = $(DFLAGS)
AB> # For IBM xlf, FDFLAGS is the same as DFLAGS with separating commas
AB> 
AB> DFLAGS         =  -D__INTEL -D__FFTW -D__USE_INTERNAL_FFTW

for parallel compilation there should be a -D__PARA here.
so you basically compiled the serial version (using mpif77 instead
of ifort doesn't automatically mean that you compile the parallel
version). so please do a 'make clean' and then re-configure with:

env MPIF90=mpif77 ./configure 

check the output of the configure command, and then re-compile.

[...]

AB> my computer run both of cpus but he didn't share kpoint he only do the same
AB> job with both cpus.

yep. this is what is supposed to happen if you start a non-parallel
program aross multiple cpus with MPI. MPI parallel programs are 
supposed to be independent until they call MPI_INIT and after
MPI_FINALIZE. a serial program (even if compiled with an MPI 
compiler wrapper) does not contain these calls and hence you get
the behavior you see and the warning from mpirun. if you want to
learn more details about this, i recommend you look into some of
the MPI tutorials on the web, that make this very clear.

cheers,
   axel.

AB> please help me i really confused why it happended.
AB> sincerely yours
AB> 
AB> 
[...]

-- 
=======================================================================
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.