[Q-e-developers] Need help from a confi-guru

[Filippo Spiga]

Dear Paolo,

there were troubles using "--enable-openmp", I've committed a fix that it
seems to work. I am trying several version of Intel and MKL on different
machines because there is still something weird.... For example I want to
trace the reason why this happens:

The following libraries have been found:
  BLAS_LIBS=  -lmkl_intel_lp64  -lmkl_intel -lmkl_core
  LAPACK_LIBS=/home/fs395/me/QE/espresso/lapack-3.2/lapack.a
  FFT_LIBS=

If you are using MKL and Intel, you can avoid
"/home/fs395/me/QE/espresso/lapack-3.2/lapack.a" at all. Am I wrong?

Moreover also this happen when you use MKL and GNU compilers:

The following libraries have been found:
  BLAS_LIBS=  -lmkl_gf_lp64  -lmkl_gnu_thread -lmkl_core
  LAPACK_LIBS=
  FFT_LIBS=

Why in this case LAPACK_LIBS is empty? If you do not mind, I will try to
test several combinations of:
- "--enable-openmp" versus "--disable-openmp"
- MKL and Intel versus MKL and GNU
- serial MKL and multi-threaded MKL (by the way, you cannot install MKL has
both sequential and multi-threaded libraries, Intel does not release those
separately)

On the Darmin cluster here at Cambridge I can basically switch between
several versions of Intel (from 10.0.023 to 13.1.0.146) and several
versions of MKL (from 10.0.3.020 and 11.0.2.146) using module. Trying
several of them just requires some time.

I like the "-mkl=..." stuff but in order to maintain backward compatibility
with old Intel compilers (previous 13.x.x or previous 12.x.x) we might need
to check also the compiler version. Is it worth the additional lines in the
configure? At the end of the day

Cheers,
FIlippo


--
Mr. Filippo SPIGA, M.Sc.
http://filippospiga.me ~ skype: filippo.spiga

«Nobody will drive us out of Cantor's paradise.» ~ David Hilbert



On Fri, Apr 5, 2013 at 3:54 PM, Paolo Giannozzi <paolo.giannozzi at uniud.it>wrote:

> On Fri, 2013-04-05 at 14:21 +0200, Lorenzo Paulatto wrote:
>
> > Can you try to use "-mkl=cluster" (scalapack) and "-mkl=sequential"?
>
> it works only for ifort, I guess. Thanks for the info.
> I wrote a dumb version of configure.ac and it seems to work.
>
> P.
> --
>  Paolo Giannozzi, Dept. Chemistry&Physics&Environment,
>  Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
>  Phone +39-0432-558216, fax +39-0432-558222
>
> _______________________________________________
> Q-e-developers mailing list
> Q-e-developers at qe-forge.org
> http://qe-forge.org/mailman/listinfo/q-e-developers
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/developers/attachments/20130408/b2ee3865/attachment.html>