[Wien] wien2k installation problem on IBM Power 7

Oliver Albertini ora at georgetown.edu
Thu Oct 3 20:50:40 CEST 2013


Hello Jianxin,

The qmix8.F error was still present when I compiled version 13 on AIX 7.1.
I also changed these manually.

With version 14 of xlf90, I did not see the loop error that you mentioned,
however, I did see it with version 11 of xlf90 on AIX 6.1. I did not use
-qalign or -qstrict. Maybe try without these? Also I recommend using the
most aggressive optimization, O5. Testing the benchmark in serial it seems
to have a significant benefit.

The '&' issue is still present in W2k13 for the SRC_hf files (several). I
added these manually.

There are several other issues getting W2k to run in the AIX environment. A
standard linux bc command and sleep command, for example. These can be
installed with rpms. There is an issue with ESSL/Lapack naming of libraries
that affects the mixer, so you should link lapack first (for the mixer
only).

Oliver


On Wed, Jul 31, 2013 at 9:52 PM, Gavin Abo <gsabo at crimson.ua.edu> wrote:

>  Try changing
>
> (/1:nr/)
>
> to
>
> (/ (i,i=1,nr) /)
>
> I believe the later is a more standard equivalent syntax.
>
>
> On 7/31/2013 9:15 PM, Zhu, Jianxin wrote:
>
> Hi Peter and Wien2k Users,
>
>  When I install wien2k on a IBM Power 7 unit with xlf90_r etc., I see the
> following compilation errors ---
>
>  1. In the more recent version like 12.1, a qmix8.F and SRC_hf folder are
> added to the package
>
>  For qmix8.F, the error comes at lines like
>
>                 write(21,2121),XUSE,' due to touching spheres'
>
>  So I manually changed it to
>                 write(21,2121) XUSE,' due to touching spheres'
>
>  2. In the subdirectory of SRC_hf, the symptom of the error is
>
>  xlf90_r -O3 -qalign -qstrict -q64 -qarch=auto -c calc_rhovalvxsl_tmp_.F
> "calc_rhovalvxsl_tmp_.F", line 25.48: 1515-019 (S) Syntax is incorrect.
>
> corresponding to the source line
>
>          r(1:nr) = r0(iat)*exp(dx(iat)*(dble((/1:nr/))-1d0))
>
>  I know that this way of coding is quite simpler but is not necessarily
> application to all compilers.
> For IBM XLF, can you suggest some option to recognize the above coding
> style?
> Otherwise, I will have to change it manually to a do-loop.
>
>  Also, for the continuation of one line, for some executable statements,
> I need to provide a second "&" sign like
>
>          if ((nonself .eqv. .false.) .and. (diaghf .eqv. .false.)) &
>            write(21,'(":CCESUM    : ",1x,f22.9,3X,"(Ry)")') ccesum
>
>  --->
>
>          if ((nonself .eqv. .false.) .and. (diaghf .eqv. .false.)) &
>        &    write(21,'(":CCESUM    : ",1x,f22.9,3X,"(Ry)")') ccesum
>
>  There are a lot of places ,where I have to do this.
> Is there a simple option for me to avoid manually changing the code?
>
>  Thanks for sharing your experience.
>
>  Cheers,
>
>  Jianxin
>
>
>
> _______________________________________________
> 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
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20131003/4d0a8e08/attachment.htm>


More information about the Wien mailing list