[Wien] 13 compilation with openmp

Laurence Marks L-marks at northwestern.edu
Mon Jul 22 14:36:37 CEST 2013


Wien2k uses lapack/blas for multithreading not openmp. To my knowledge
there is no openmp used anywhere. I cannot find any openmp directives
in lapwso beyond the !$ DOACROSS directive in lap_bp.f. I believe this
is a perhaps slightly customized version of a lapack command, and the
$! DOACROSS directive is an obsolete remnant. The code compiles fine
if you do not use "-openmp" in the compilation line. I suggest
changing this line to !!$ DOACROSS and not using -openmp. I have not
checked the other cases but I suspect it is similar.


On Mon, Jul 22, 2013 at 6:41 AM, Torsten Weissbach
<torsten.weissbach at physik.tu-freiberg.de> wrote:
> Hello,
>
> I have some questions about the compilation of WIEN2k 13 with openmp and
> FFTW support. I'm using ifort 11.1 on Scientific Linux together with the
> mkl 10.3.9 (from a later compiler) and fftw3. The compilation functions
> when I add -lifcoremt -lirc to the LDFLAGS (there was a strange error
> occurring before) and results in a working threaded program.
>
> But only if the FOPT string does not contain -openmp. Instead -openmp is
> added to the R_LIBS. If I set -openmp in FOPT, the I get compiler errors
> in hf:
>
> compile.msg:
>
> ........snip......
>
> ifort -FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -traceback -O3
> -xHost -mc
> model=medium
> -I/cm/shared/apps/intel-cs/composer_xe_2011_sp1.9.293/mkl/include -
> openmp -DFFTW3 -I/home/weissbach/fftw/include -c calc_h_tmp_.F
> : catastrophic error: **Internal compiler error: segmentation violation
> signal r
> aised** Please report this error along with the circumstances in which
> it occurr
> ed in a Software Problem Report.  Note: File and line given may not be
> explicit cause of this error.
>
> compilation aborted for calc_h_tmp_.F (code 3)
> make[1]: *** [calc_h.o] Error 3
> make[1]: Leaving directory `/home/weissbach/WIEN2k_13_omptest/SRC_hf'
> make: *** [real] Error 2
>
> ........snip.............
> (same for the complex version)
>
>
> and lapwso:
>
> compile.msg:
>
> ........snip......
>
> ifort  -FR -mp1 -w -prec_div -pc80 -pad -ip -DINTEL_VML -traceback -O3
> -xHost -mcmodel=medium
> -I/cm/shared/apps/intel-cs/composer_xe_2011_sp1.9.293/mkl/include
> -openmp -DFFTW3 -I/home/weissbach/fftw/include -c lap_bp.f
> lap_bp.f(575): error #5082: Syntax error, found IDENTIFIER 'LOCAL' when
> expecting one of: ( % : . = =>
> !$ DOACROSS Local (KP,KK,KKB,KKK,KKKB,KTMP1,KTMP2,KTMP3)
> ------------^
> lap_bp.f(4656): error #5082: Syntax error, found IDENTIFIER 'LOCAL' when
> expecting one of: ( % : . = =>
> !$ DOACROSS Local (KP,KK,KKB,KKK,KKKB,KTMP1,KTMP2,KTMP3)
> ------------^
> compilation aborted for lap_bp.f (code 1)
> make: *** [lap_bp.o] Error 1
>
> ........snip......
>
>
> You see that I also included an external fftw hoping to get openmp
> parallel fft. However it seems that lapw0 (when FOPT does not contain
> -openmp) is not threaded.
>
> It would also be very helpful to know some typical testcases that will
> show whether the parts of the program are running well. Can you suggest
> some?
>
> Best regards,
> Torsten
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:  http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html



-- 
Professor Laurence Marks
Department of Materials Science and Engineering
Northwestern University
www.numis.northwestern.edu 1-847-491-3996
"Research is to see what everybody else has seen, and to think what
nobody else has thought"
Albert Szent-Gyorgi


More information about the Wien mailing list