[Pw_forum] newbie - installation problem

Axel Kohlmeyer akohlmey at cmm.chem.upenn.edu
Tue Mar 3 17:32:00 CET 2009

On Tue, 3 Mar 2009, Carlo Nervi wrote:

CN> Dear Lorenzo,
CN> thank you very much for quick reply.
CN> Indeed you perfectly get the problem.
CN> I still have some difficulties, (how to correctly set the flags for SMP 
CN> parallel compilation - no MPI, even though I have MPI installed) but I'm 
CN> working on them.
CN> Indeed I have a 8 CPU SMP linux (no cluster, a simple single dual socket 
CN> xeon MB).

you still want to run MPI on it. it is much more efficient
in all but a few strange corner cases, and you want to make
certain that OpenMP is disabled in that case.


CN> Thank you again,
CN>    Carlo
CN> Lorenzo Paulatto ha scritto:
CN> > On Tue, 03 Mar 2009 15:44:12 +0100, Carlo Nervi <carlo.nervi at unito.it> wrote:
CN> >> When I type make all, ifort compile all, but it exit with the following
CN> >> message:
CN> >> ipo: warning #11041: unresolved iotk_print_kinds_x_
CN> >>          Referenced in /tmp/ipo_ifort86WxDo.o
CN> > 
CN> > Dear carlo,
CN> > QE configure script uses the default system archiver (ar) when using ifort compiler instead of intel's own wrapper (xiar); unluckily interprocedural optimization (IPO) does not work with ar.
CN> > 
CN> > You have two possible solutions:
CN> > 1. disable IPO (by setting the -noipo in F90FLAGS in file make.sys)
CN> > 2. use xiar instead of ar (by setting the AR variable to xiar in file make.sys)
CN> > 
CN> > Personally I don't think IPO gives a substantial performance increase, hence I just disable it most of the time (furthermore it's disabled by default on most ifort versions).
CN> > 
CN> > cheers
CN> > 
CN> > 

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 mailing list