[MITgcm-support] relocation truncated to fit: R_386_32

Martin Losch mlosch at awi-bremerhaven.de
Mon Aug 7 03:41:51 EDT 2006


Hi Nicolas,

I use -mcmodel=medium with the pgi compiler and it works fine, but  
you should put it in your FFLAGS (not the CPPFLAGS).

Martin

On Jul 28, 2006, at 5:19 PM, Nicolas Wienders wrote:

>
>
>>
>> This happens because your matrices end up larger than 2GB in size  
>> per process.
>> Either compile with the medium memory model (-mmodel=medium) or  
>> compile
>> everything with -KPIC. The former may require the MPI library to be
>> recompiled with -mmodel=medium.
>
> Thanks Constantinos,
> I tried the second option but i got the same problems.
> I must not be using the flag properly i am sure.
> Here is my opt file:
>
>
> FC=mpif77
> CC=mpicc
> DEFINES='-D_BYTESWAPIO -DWORDLENGTH=4'
> CPP='cpp  -traditional -KPIC'
> NOOPTFLAGS='-O0'
> #MAKEDEPEND=/usr/X11R6/bin/gccmakedep
> #MAKEDEPEND=/usr/X11R6/bin/makedepend
> LIBS='-L/usr/lib -lnetcdf'
> INCLUDES='-I/usr/lib/mpich/include'
>
>
> #  For IEEE, use the "-ffloat-store" option
> if test "x$IEEE" = x ; then
>     FFLAGS='-Wimplicit -Wunused -Wuninitialized'
>     FOPTIM='-O3 -funroll-loops'
> else
>     FFLAGS='-Wimplicit -Wunused -ffloat-store'
>     FOPTIM='-O0 '
> fi
>
>
> Thanks. n
>
>
>
>
>
>
>
> --
>
> Nicolas Wienders                   (850) 644-1987 (office)
> Oceanography - OSB415                294-3542 (cell)
> Florida State University                    644-2581 (fax)
> Tallahassee, FL - 32306    wienders at ocean.fsu.edu
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list