[Wien] Questions in goto library on Itanium2
Jorissen Kevin
Kevin.Jorissen at ua.ac.be
Tue Jan 6 17:30:09 CET 2004
Hi, I reversed the order of the libraries when linking (-lgoto... -llapack_lapw) (might be relevant, as the goto libraries also contain some (though not all) lapack routines).
Do you get the same errors when using no optimization at all (-O0) ?
I remember there's some issue about a routine xerbla for the goto library (see website); did you check that?
Good luck,
Kevin Jorissen.
-----Oorspronkelijk bericht-----
Van: Shunli Shang [mailto:S.Shang at tnw.tudelft.nl]
Verzonden: di 1/6/2004 3:45
Aan: wien at zeus.theochem.tuwien.ac.at
CC:
Onderwerp: [Wien] Questions in goto library on Itanium2
Dear Wien2k Users,
I met troubles when I ran wien2k using the (efc7.1 + goto library ) on
Itanium2 processors. That is the wien2k is running well, but the results are
incorrect. E.g. for the demo case TiC, after 20 cycles the energy or the
charge distance are still not convergent. In fact the results of the first
cycle of TiC case are wrong, the results of the first cycle of TiC are as
fellows (efc7.1 + goto library):
ENE: -1789.755269 (the right one should be: -1784.12)
FER: 0.52773 (the right one should be: 0.73)
DIS: 0.4593901 (the right one should be: 0.857)
Details of the supercomputer are: SGI Altix 3700 system with Intel Itanium2
(1.3GHz) processors and Red Hat Linux operating system.
The tested speeds for the benchmark case are as follows.
efc7.1 + mkl6.0 : 298 sec
efc7.1 + goto library : 209 sec
Wien2k results for (efc7.1 + mkl6.0) are correct, no problem. The compiling
options for (efc7.1 + goto library) are as follows:
O: -FR -mp -w -O2
L: -L../SRC_lib -Vaxlib
P: 'DParallel'
R: -llapark_lapw -lgoto_it2-r0.9
Is someone using the Itanium2 processor and goto library? Could you please
tell me how to get the correct results?
Best Regards,
Shunli Shang
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 5658 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20040106/179339d4/attachment.bin
More information about the Wien
mailing list