[MITgcm-support] Compiler error when doubling horizontal resolution

[Roland Young]

Hi Habib,

Thanks for your reply. I had seen a few of those links before - I found 
a few tips to get around the Fortran problem using information from 
those links, but I was unable to find anything that fixed the C problem.

In any case, I was able to get around the problem in practice by 
increasing the number of cores. I was trying to test it on a 16-core 
desktop (before running it on 128 or 256). When I increased the number 
of cores to 64 for my test run the size of the arrays for each process 
became small enough that the requirement for mcmodel=medium went away.

Thanks,

Roland





On 23/12/15 20:57, Habib Toye Mahamadou Kele wrote:
> Hi Roland,
>
> The issue you encountered is due to the size of the arrays (you
> increased the resolution therefore the arrays became larger). So you
> need to review the memory storage management, by using compiler flag
> (what you are trying to do) or modify the code (what I think you do not
> want to do).
>
> I did not try to compile the code to reproduce your error and try to
> solve it (because it would take me some time to set the code and compile
> it).
>
> However the following links might be helpfull:
>
> http://mitgcm.org/pipermail/mitgcm-support/2006-September/004231.html
>
> http://mitgcm.org/pipermail/mitgcm-support/2006-September/004232.html
>
>
> https://software.intel.com/en-us/forums/intel-fortran-compiler-for-linux-and-mac-os-x/topic/270056?db=5
> and then
> https://software.intel.com/en-us/articles/avoiding-relocation-errors-when-building-applications-with-large-global-or-static-data-on-intel64/
>
>
> http://stackoverflow.com/questions/10486116/what-does-this-gcc-error-relocation-truncated-to-fit-mean
> (read the ticked response)
> and then
> https://www.technovelty.org/c/relocation-truncated-to-fit-wtf.html
>
>
> http://stackoverflow.com/questions/12916176/gfortran-for-dummies-what-does-mcmodel-medium-do-exactly
>
>
> http://stackoverflow.com/questions/11131998/from-which-version-of-gcc-does-mcmodel-large-work
>
>
> Furthermore,
> "the -shared-intel flag is not recognized by gcc" since it is a specific
> intel flag.
>
>
> Regards,
> Habib TOYE
>
>
> On Wed, Dec 23, 2015 at 9:43 PM, Roland Young
> <Roland.Young at physics.ox.ac.uk <mailto:Roland.Young at physics.ox.ac.uk>>
> wrote:
>
>     Hi,
>
>     I have just increased the horizontal resolution of my model from
>     512x256x33 to 1024x512x33, but I am getting the following compiler
>     error (example given for a C routine, but similar for Fortran):
>
>     /home/jupiter/gfd/young/runs/testing/h222_test/build/sigreg.c:34:(.text+0x3):
>     relocation truncated to fit: R_X86_64_PC32 against symbol
>     `sigreg_ip' defined in COMMON section in sigreg.o
>
>     I managed to remove this error from the Fortran routines by adding
>     the following compiler flags
>
>     FFLAGS="$FFLAGS -mcmodel=medium -shared-intel”
>
>     However, I have not been able to remove it from the C routines. I
>     have tried
>
>     CFLAGS="$CFLAGS -mcmodel=medium”
>     CFLAGS="$CFLAGS -mcmodel=large”
>     CFLAGS="$CFLAGS -mmodel=medium”
>     CFLAGS="$CFLAGS -mmodel=large”
>
>     and all give the same error (the -shared-intel flag is not
>     recognised by gcc). I think I’m on the right lines but what else
>     needs to be done to fix the C compiler error?
>
>     Thanks,
>
>     Roland
>
>
>     _______________________________________________
>     MITgcm-support mailing list
>     MITgcm-support at mitgcm.org <mailto:MITgcm-support at mitgcm.org>
>     http://mitgcm.org/mailman/listinfo/mitgcm-support
>
>
>
> ------------------------------------------------------------------------
> This message and its contents, including attachments are intended solely
> for the original recipient. If you are not the intended recipient or
> have received this message in error, please notify me immediately and
> delete this message from your computer system. Any unauthorized use or
> distribution is prohibited. Please consider the environment before
> printing this email.
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
>


-- 
Dr Roland Young, Postdoctoral Research Assistant
Department of Physics, University of Oxford
t: +44 (0)1865 272337 | e: young at atm.ox.ac.uk
w: https://www.physics.ox.ac.uk/contacts/people/youngr