[Wien] clmsum - still no luck

[Michael Gurnett]

I added -fpe0 -ftrapuv -traceback -align (I was already using -pad)

Got the following now instead

mixer              0000000000403CA9  Unknown               Unknown
Unknown
libc.so.6          0000003781C1C40F  Unknown               Unknown  Unknown
mixer              0000000000403D6A  Unknown               Unknown  Unknown
mixer              000000000040C614  Unknown               Unknown  Unknown
mixer              00000000004177B9  normal_                    36  normal.f
Image              PC                Routine            Line        Source             
forrtl: error (73): floating divide by zero

The crash usually occurs in lapw0 (both single and parallel mode) so
this is new (guessing it has to do with the new flags catching this).
The kernel is the absolute latest fedora has to offer so no sleeping nfs
bug (unless they reintroduced it).

Michael

On Sat, 2006-04-22 at 07:56 -0500, L. D. Marks wrote:
> Some suggestions, probably useless:
> 1) compile with -ftrapuv -fce0 -ftraceback. These should not slow things 
> down, but may reveal subtle coding problems that get hidden.
> 2) Try both modes -cxxlib-gcc and -cxxlib-icc
> 3) Test if you get the same problems using lapw0_mpi as with lapw0. Does 
> it also occur when running lapw0 in single remote mode?
> 4) Add -check to the compilation options, and rerun (being patient as it 
> will take somewhat longer).
> 5) Try adding -align -pad. I see no reason why these should have any 
> effect, but....
> 6) Check that you don't have the asleep NFS problem. This one is like a 
> trojan in that it does not obviously show itself.
> 
> On Sat, 22 Apr 2006, Michael Gurnett wrote:
> 
> > I've been trying now for some time to get this problem of the zero
> > columns to be fixed, without any luck
> >
> > I've tried the default setting as you recommended Peter, but I have to
> > remove the static-libcxa switch as that would not compile (but this is
> > not required if I understand Gerhard correctly). So with all other
> > setting exactly as default I compile using fce 9.0 and mkl 8.0.1 (have
> > tried both cc and icc for the c compiler) on fedora 4. This compiles
> > fine but the problem is their. In fact for every type of compile I have
> > done that has compiled with zero errors, the problem has been the same.
> > I've just downloaded the wien2k executables to see what they do, but I'm
> > guessing that these are compiled for 32 bit and I know they are going to
> > be slow if they work.
> >
> > Kevin, did you ever get your code to recompile and work properly?
> >
> > Really hope someone can help me with this
> >
> > Michael
> >
> >
> > _______________________________________________
> > Wien mailing list
> > Wien at zeus.theochem.tuwien.ac.at
> > http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> >
> 
> Note: if you have an old email address for me, please note that "nwu" has
> been changed to "northwestern".
> -----------------------------------------------
> Laurence Marks
> Department of Materials Science and Engineering
> MSE Rm 2036 Cook Hall
> 2225 N Campus Drive
> Northwestern University
> Evanston, IL 60208, USA
> Tel: (847) 491-3996 Fax: (847) 491-7820
> email: L-marks at northwestern dot edu
> http://www.numis.northwestern.edu
> -----------------------------------------------
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>