[MITgcm-support] [EXTERNAL] Re: Compiling Problems

Samar Khatiwala samar.khatiwala at earth.ox.ac.uk
Tue Aug 20 17:50:21 EDT 2024


Hi Madeleine,

It doesn’t seem you're using a derecho-specific optfile when invoking genmake. Nor do you appear to be specifying the path to the MPI library.

If I remember, on derecho the compiler wrappers mpif90 etc don’t automatically link to the MPI libraries. You need to explicitly specify the path in your optfile. You can find out what libraries/include paths are being used by doing 'mpif90 -v’ (or 'mpif90 -show' on sensible MPI installations which unfortunately cray isn’t).

Samar

On Aug 20, 2024, at 9:50 PM, Madeleine Youngs <myoungs at umd.edu> wrote:

Hi All,

I have done some digging on my own.  I ran testreport, which seemed to work fine with a single processor, but seems to be having similar issues for building with MPI.  I thought to do this so it wasn’t dependent on my particular experiement.   For example:

Experiment:  adjustment.128x64x1

genmake ... successful
clean build-dir: make Clean ...
successful
make depend ...
successful
make  ...
all_proc_die.f(789): error #5082: Syntax error, found END-OF-FILE when expecting one of: <LABEL> <END-OF-STATEMENT> ; <IDENTIFIER> TYPE MODULE ELEMENTAL IMPURE NON_RECURSIVE ...
C--   Include MPI standard Fortran header file
----------------------------------------------^
all_proc_die.f(270): error #6404: This name does not have a type, and must have an explicit type.   [MYTHARG]
      SUBROUTINE ALL_PROC_DIE( myThArg )
-------------------------------^
compilation aborted for all_proc_die.f (code 1)
make[1]: *** [Makefile:1223: all_proc_die.o] Error 1
make[1]: Leaving directory '/glade/derecho/scratch/myoungs/MITgcm/verification/adjustment.128x64x1/build'
make: *** [Makefile:1135: fwd_exe_target] Error 2
failed


On Aug 14, 2024, at 11:51 AM, Madeleine Youngs <myoungs at umd.edu> wrote:

Hi All,

Thanks for all the support! I have tried the make CLEAN and the gfort compiler and I get the same error.  I do not have any code mods so it can’t be that.  I did successfully create diagnostics_check.f:

Would you want me to attach a file for the .f and .F files?

I also noticed this:
original 'Makefile' generated but was unable to
   run compiled test-program (please see 'genmake.log’)


Thanks,
Madeleine

running: check_fortran_compiler
 ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -c genmake_tcomp.f
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
 ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -o genmake_tcomp genmake_tcomp.o
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
 mpirun -np 1 ./genmake_tcomp
No host list provided
 --> set FC_CHECK= 3 /5

running: look_for_makedepend()
 --> set MAKEDEPEND=$(TOOLSDIR)/xmakedepend

running: check_HAVE_CLOC()
 running: get_fortran_c_namemangling()
 cc -O2 -ip -fPIC -c genmake_test.c
icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is deprecated and will be removed from product release in the second half of 2023. The Intel(R) oneAPI DPC++/C++ Compiler (ICX) is the recommended compiler moving forward. Please transition to use this compiler. Use '-diag-disable=10441' to disable this message.
genmake_test.c(1): warning #266: function "tsub" declared implicitly
  void tcall( char * string ) { tsub( string ); }
                                ^

 ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -c genmake_tcomp.f
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
 --> set FC_NAMEMANGLE='#define FC_NAMEMANGLE(X)  X ## _'
cc -O2 -ip -fPIC -c genmake_tc_1.c
icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is deprecated and will be removed from product release in the second half of 2023. The Intel(R) oneAPI DPC++/C++ Compiler (ICX) is the recommended compiler moving forward. Please transition to use this compiler. Use '-diag-disable=10441' to disable this message.
ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -o genmake_tc genmake_tc_2.f genmake_tc_1.o
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
  HELLO WORLD   1723660750.37632
 --> set HAVE_CLOC='t'

running: check_HAVE_SETRLSTK()
cc -O2 -ip -fPIC -c genmake_tc_1.c
icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is deprecated and will be removed from product release in the second half of 2023. The Intel(R) oneAPI DPC++/C++ Compiler (ICX) is the recommended compiler moving forward. Please transition to use this compiler. Use '-diag-disable=10441' to disable this message.
      program hello
      external setrlstk
      call setrlstk()
      end
ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -o genmake_tc genmake_tc_2.f genmake_tc_1.o
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
 --> set HAVE_SETRLSTK='t'

running: check_HAVE_SIGREG()
cc -O2 -ip -fPIC -c genmake_tc_1.c
icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is deprecated and will be removed from product release in the second half of 2023. The Intel(R) oneAPI DPC++/C++ Compiler (ICX) is the recommended compiler moving forward. Please transition to use this compiler. Use '-diag-disable=10441' to disable this message.
genmake_tc_1.c(11): warning #274: declaration is not visible outside of function
      unsigned int sn, siginfo_t  si, struct ucontext *sc )
                                             ^

      program hello
      integer anint
      common /iv/ anint
      external sigreg
      call sigreg(anint)
      end
ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -o genmake_tc genmake_tc_2.f genmake_tc_1.o
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
 --> set HAVE_SIGREG='t'

running: check_HAVE_STAT()
cc -O2 -ip -fPIC -c genmake_tc_1.c
      program hello
      integer nbyte
      call tfsize(nbyte)
      print *," HELLO WORLD", nbyte
      end
ftn -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -o genmake_tc genmake_tc_2.f genmake_tc_1.o
 --> set HAVE_STAT='t'

running: check_netcdf_libs()
===  genmake_tnc.F  >>>
      program fgennc
#include "netcdf.inc"
      integer iret, ncid, xid
      iret = nf_create('genmake_tnc.nc', NF_CLOBBER, ncid)
      IF (iret .NE. NF_NOERR) write(*,*) NF_STRERROR(iret)
      iret = nf_def_dim(ncid, 'X', 11, xid)
      IF (iret .NE. NF_NOERR) write(*,*) NF_STRERROR(iret)
      iret = nf_close(ncid)
      IF (iret .NE. NF_NOERR) write(*,*) NF_STRERROR(iret)
      end
<<<  genmake_tnc.F  ===

cat genmake_tnc.F |  cpp -P -traditional -DALLOW_USE_MPI -DALWAYS_USE_MPI -DWORDLENGTH=4 -DNML_TERMINATOR -DALLOW_USE_MPI -DHAVE_SYSTEM -DHAVE_FDATE -DHAVE_ETIME_FCT -DHAVE_CLOC -DHAVE_SETRLSTK -DHAVE_SIGREG -DHAVE_STAT -I/opt/cray/pe/mpich/8.1.25/ofi/intel/19.0/include/
  WARNING: CPP failed to pre-process the netcdf test.
    Please check that $INCLUDES is properly set.
ftn   -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -c genmake_tnc.f  \
  &&  ftn   -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -o genmake_tnc.o
ifort: command line warning #10121: overriding '-march=core-avx2' with '-march=core-avx2'
genmake_tnc.f(12): error #5082: Syntax error, found END-OF-FILE when expecting one of: <LABEL> <END-OF-STATEMENT> ; <IDENTIFIER> TYPE MODULE ELEMENTAL IMPURE NON_RECURSIVE ...
      program fgennc
--------------------^
compilation aborted for genmake_tnc.f (code 1)

==> try again with added '-lnetcdf'
cat genmake_tnc.F |  cpp -P -traditional -DALLOW_USE_MPI -DALWAYS_USE_MPI -DWORDLENGTH=4 -DNML_TERMINATOR -DALLOW_USE_MPI -DHAVE_SYSTEM -DHAVE_FDATE -DHAVE_ETIME_FCT -DHAVE_CLOC -DHAVE_SETRLSTK -DHAVE_SIGREG -DHAVE_STAT -I/opt/cray/pe/mpich/8.1.25/ofi/intel/19.0/include/ > genmake_tnc.f \
 &&  ftn   -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -c genmake_tnc.f \
 &&  ftn   -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -o genmake_tnc genmake_tnc.o  -lnetcdf

==> try again with added '-lnetcdff -lnetcdf'
cat genmake_tnc.F |  cpp -P -traditional -DALLOW_USE_MPI -DALWAYS_USE_MPI -DWORDLENGTH=4 -DNML_TERMINATOR -DALLOW_USE_MPI -DHAVE_SYSTEM -DHAVE_FDATE -DHAVE_ETIME_FCT -DHAVE_CLOC -DHAVE_SETRLSTK -DHAVE_SIGREG -DHAVE_STAT -I/opt/cray/pe/mpich/8.1.25/ofi/intel/19.0/include/ > genmake_tnc.f \
 &&  ftn   -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -c genmake_tnc.f \
 &&  ftn   -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g -O2 -align -ip -o genmake_tnc genmake_tnc.o  -lnetcdf
 --> set HAVE_NETCDF=''

running: check_HAVE_FLUSH()
  check_HAVE_FLUSH: successful
 --> set HAVE_FLUSH='t'





On Aug 14, 2024, at 10:14 AM, Jody Klymak <jklymak at uvic.ca> wrote:

Hi Madeleine

When I have linking problems like this I sometimes fine `make CLEAN` to be a useful step if I’m not starting from a clean install.

Just a shot in the dark.

Cheers,  Jody

On Aug 14, 2024, at 10:05 AM, Wang, Ou (US 329B) <ou.wang at jpl.nasa.gov> wrote:

You don't often get email from ou.wang at jpl.nasa.gov<mailto:ou.wang at jpl.nasa.gov>. Learn why this is important<https://aka.ms/LearnAboutSenderIdentification>
Hi Madeleine,

Can you share your version of diagnostics_ini_io.f and diagnostics_ini_io.F? I would think the error is related to a broken link to “DIAGNOSTICS.h”, not something related to the opt file. Also, did the compilation successfully generate diagnostics_check.f, which uses the variable “diagMdsDir” as well?
It would be helpful if you can also share your versions of diagnostics_check.f and diagnostics_check.F.

Best,
Ou Wang


From: MITgcm-support <mitgcm-support-bounces at mitgcm.org<mailto:mitgcm-support-bounces at mitgcm.org>> on behalf of Madeleine Youngs <myoungs at umd.edu<mailto:myoungs at umd.edu>>
Date: Tuesday, August 13, 2024 at 8:08 PM
To: mitgcm-support at mitgcm.org<mailto:mitgcm-support at mitgcm.org> <mitgcm-support at mitgcm.org<mailto:mitgcm-support at mitgcm.org>>
Subject: [EXTERNAL] Re: [MITgcm-support] Compiling Problems
Hi Takaya,

That didn’t seem to fix the issue, but thank you for sending this along!

Madeleine


On Aug 13, 2024, at 6:10 PM, Takaya Uchida <tu22 at fsu.edu<mailto:tu22 at fsu.edu>> wrote:

Hi Madeleine,

I’ve copy and pasted below the opt file I ended up using to compile MITgcm on Derecho.

#
# Build options for Derecho at NCAR
#

FC=ftn
CC=cc
DEFINES='-DALLOW_USE_MPI -DALWAYS_USE_MPI -DWORDLENGTH=4 -DNML_TERMINATOR'
CPP='cpp -P -traditional'
EXTENDED_SRC_FLAG='-132'
NOOPTFLAGS='-O2 -g -march=core-avx2'
LIBS=''
INCLUDES='-I/opt/cray/pe/mpich/8.1.25/ofi/intel/19.0/include/'

if test "x$DEVEL" != x ; then
    FFLAGS="$FFLAGS  -W0 -WB -convert big_endian -assume byterecl -fPIC -O0 -g -noalign -fpstkchk -check all -fpe0 -traceback -ftrapuv -fpmodel except -warn all -march=core-avx2"
else
    FFLAGS="$FFLAGS  -W0 -WB -convert big_endian -assume byterecl -fPIC -march=core-avx2 -g"
fi

if test "x$IEEE" = x ; then
        FOPTIM='-O2 -align -ip'
        else
        FOPTIM='-O2 -align -ip -mp'
fi
CFLAGS='-O2 -ip -fPIC'

The list of modules I load are:

module load ncarenv/23.09 craype/2.7.23 intel-classic/2023.2.1 ncarcompilers/1.0.0 cray-mpich/8.1.27 hdf5/1.12.2 netcdf/4.9.2

----------------------------------
Takaya Uchida (内田貴也)
Assistant Research Scientist
Florida State University
Center for Ocean-Atmospheric Prediction Studies
https://roxyboy.github.io/<https://urldefense.us/v3/__https:/roxyboy.github.io/__;!!PvBDto6Hs4WbVuu7!NnRx5vlbDYZWIRraW2gZnWzzXsE782jeSZ_0N-DP52xAVkHbj-uEwC3rZ_U5PHL8Ay0r9294n_QD_ImbsRJV$>

<Center for Ocean-Atmospheric Prediction Studies signature alt.png>


On Aug 13, 2024, at 8:51 PM, Madeleine Youngs <myoungs at umd.edu<mailto:myoungs at umd.edu>> wrote:

Hi Everyone,

I am working on compiling checkpoint68y on Derecho using no opt file (?) and I am getting this error.  Any advice on what the issue is?


mpif90 -W0 -WB -msse2 -convert big_endian -assume byterecl -O2 -align -lmpich -lnetcdff -lnetcdf  -c diagnostics_fract_fill.f
cat diagnostics_ini_io.F |  cpp  -traditional -P -DALLOW_USE_MPI -DALWAYS_USE_MPI -DWORDLENGTH=4 -DALLOW_USE_MPI -DHAVE_SYSTEM -DHAVE_FDATE -DHAVE_ETIME_FCT -DHAVE_CLOC -DHAVE_SETRLSTK -DHAVE_SIGREG -DHAVE_STAT -DHAVE_FLUSH  -I.  | ../../../tools/set64bitConst.sh  > diagnostics_ini_io.f
mpif90 -W0 -WB -msse2 -convert big_endian -assume byterecl -O2 -align -lmpich -lnetcdff -lnetcdf  -c diagnostics_ini_io.f
Cannot disable Fortran error message 6404
diagnostics_ini_io.f(2005): error #6404: This name does not have a type, and must have an explicit type.   [DIAGMDSDIR]
      IF ( diag_mdsio .AND. (diagMdsDir.NE.' ') .AND.
-----------------------------^
Cannot disable Fortran error message 6404
diagnostics_ini_io.f(2006): error #6404: This name does not have a type, and must have an explicit type.   [DIAGMDSDIRCREATE]
     &     diagMdsDirCreate .AND. (mdsioLocalDir.EQ.' ') ) THEN
-----------^
Cannot disable Fortran error message 6514
diagnostics_ini_io.f(2009): error #6514: Substring or array slice notation requires CHARACTER type or array.   [DIAGMDSDIR]
        WRITE(namBuf,'(3A)') ' mkdir -p ', diagMdsDir(1:iL), ' '
-------------------------------------------^
compilation aborted for diagnostics_ini_io.f (code 1)
make[1]: *** [Makefile:1776: diagnostics_ini_io.o] Error 1
make[1]: Leaving directory '/glade/derecho/scratch/myoungs/MITgcm/MultiWidthExperiments/Present9.20.16/build'
make: *** [Makefile:1688: fwd_exe_target] Error 2




Thanks,
Madeleine
_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org<mailto:MITgcm-support at mitgcm.org>
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support<https://urldefense.us/v3/__http:/mailman.mitgcm.org/mailman/listinfo/mitgcm-support__;!!PvBDto6Hs4WbVuu7!NnRx5vlbDYZWIRraW2gZnWzzXsE782jeSZ_0N-DP52xAVkHbj-uEwC3rZ_U5PHL8Ay0r9294n_QD_GmOPdGy$>

_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org<mailto:MITgcm-support at mitgcm.org>
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support<https://urldefense.us/v3/__http:/mailman.mitgcm.org/mailman/listinfo/mitgcm-support__;!!PvBDto6Hs4WbVuu7!NnRx5vlbDYZWIRraW2gZnWzzXsE782jeSZ_0N-DP52xAVkHbj-uEwC3rZ_U5PHL8Ay0r9294n_QD_GmOPdGy$>

_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org<mailto:MITgcm-support at mitgcm.org>
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support

_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support


_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org
http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20240820/44b13b06/attachment-0001.html>


More information about the MITgcm-support mailing list