[Wien] did parallel calculation take effect?

Laurence Marks L-marks at northwestern.edu
Sun Oct 12 14:49:09 CEST 2008


While we can partially help you with this, you do need to do some
thinking yourself. Some key questions:
a) Which are the "right" libraries to use? Since you appear to have
multiple copies, you need to verify what each of them are otherwise
you might be compiling with one and running with another.
b) Is LD_LIBRARY_PATH set correctly on the compute nodes? For
instance, do an ssh (or rsh) to one of them and then "echo
$LD_LIBRARY_PATH".
c) Are the shared library dependencies setup correctly? Do "man ldd",
"ldd lapw1" on your head node and also on the compute nodes. You may
want to look at "man ldconfig" as well.
d) Is LD_LIBRARY_PATH being exported in jobs? This is a slightly more
subtle issue, and you may simply want to run a job that does "echo
$LD_LIBRARY_PATH" and see the results.
e) Do you want to compile with static intel libraries or dynamic? Do
"man ifort" and/or "ifort -help link". If you don't understand the
difference between the two do a google search on "static linking" and
start reading.

On Sat, Oct 11, 2008 at 11:24 PM, vinct8954
<songwenbin2000csu at yahoo.com.cn> wrote:
> dear Peter Blaha :
> thanks for your answer, and i changed the share memory for NO。and choose rsh
> for the parallel calculation。as usual,there is no errors while compiling,but
> when i intended to run a test for parallel with -p there is another puzzled
> problems.
> lapw 1:error while loading share libraries:libmkl_lapack64.so:cannot open
> shared object file: no such file or directory.
> but when i run the same test in single calculation there is no error like
> this.
> and the LD_LIBRARY_PATH are settle properly as follow:
> LD_LIBRARY_PATH=/export/compiler/intel91/lib:/export/compiler/intel91/lib
> :/export/mathlib/cmkl81/lib/em64t:/export/compiler/intel91/lib:
> /export/home/test/magpar-0_5/libs/sundials/lib
> and my compile option is under :/export/mathlib/cmkl81/lib/em64t
>
> and i checked the other nodes for "env" and still could find the same path.
>
> i have no idear to change the setting.and how to solve this problem.
> by the way. i used "locate" can find different path of libmkl_lapack64.so in
> master nodes:
> /export/mathlib/cmkl81/lib/64/libmkl_lapack64.so
> /export/mathlib/cmkl81/lib/32/libmkl_lapack64.so
> /export/mathlib/cmkl81/lib/em64t/libmkl_lapack64.so
> /export/Abinitio/MS4.0/share/lib/libmkl_lapack64.so
> /export/Abinitio/MStudio/share/lib/libmkl_lapack64.so
>
>
> ________________________________
> 雅虎邮箱,您的终生邮箱!
>
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>
>



-- 
Laurence Marks
Department of Materials Science and Engineering
MSE Rm 2036 Cook Hall
2220 N Campus Drive
Northwestern University
Evanston, IL 60208, USA
Tel: (847) 491-3996 Fax: (847) 491-7820
email: L-marks at northwestern dot edu
Web: www.numis.northwestern.edu
Chair, Commission on Electron Crystallography of IUCR
www.numis.northwestern.edu/
Electron crystallography is the branch of science that uses electron
scattering to study the structure of matter.


More information about the Wien mailing list