-fpe0 failures

[anton-seaice]

When compiling and running with -fpe0 I get this failure during model initialisation:

forrtl: error (73): floating divide by zero
Image              PC                Routine            Line        Source             
libpthread-2.28.s  000014A4E8638990  Unknown               Unknown  Unknown
libimf.so          000014A4EABBF1B7  __libm_log10_l9       Unknown  Unknown
access-om3-MOM6-C  0000000002110CFF  time_scalar_mult          778  time_manager.F90
access-om3-MOM6-C  0000000002110ECE  time_divide               806  time_manager.F90
access-om3-MOM6-C  0000000000795F07  write_energy              496  MOM_sum_output.F90
access-om3-MOM6-C  00000000005E1279  finish_mom_initia        3775  MOM.F90
access-om3-MOM6-C  000000000057BAAA  update_ocean_mode         608  mom_ocean_model_nuopc.F90
access-om3-MOM6-C  000000000053C349  modeladvance             1979  mom_cap.F90
libesmf.so         000014A4EC274E67  execute                   377  ESMCI_MethodTable.C
libesmf.so         000014A4EC274D05  c_esmc_methodtabl         317  ESMCI_MethodTable.C
libesmf.so         000014A4EC4058BB  esmf_methodgridco        1287  ESMF_AttachMethods.F90
libesmf.so         000014A4ECB45B8E  routine_run              2212  NUOPC_ModelBase.F90
libesmf.so         000014A4EBFCFB08  callVFuncPtr             2187  ESMCI_FTable.C
libesmf.so         000014A4EBFCF54C  ESMCI_FTableCallE         844  ESMCI_FTable.C
libesmf.so         000014A4EC30EC14  enter                    1216  ESMCI_VM.C
libesmf.so         000014A4EBFD01E0  c_esmc_ftablecall        1001  ESMCI_FTable.C
libesmf.so         000014A4EC505B28  esmf_compexecute         1273  ESMF_Comp.F90
libesmf.so         000014A4EC772344  esmf_gridcomprun         1927  ESMF_GridComp.F90
libesmf.so         000014A4ECAFF8B5  routine_executegr        3698  NUOPC_Driver.F90
libesmf.so         000014A4ECB014D8  executerunsequenc        3944  NUOPC_Driver.F90
libesmf.so         000014A4EC274E67  execute                   377  ESMCI_MethodTable.C
libesmf.so         000014A4EC274D05  c_esmc_methodtabl         317  ESMCI_MethodTable.C
libesmf.so         000014A4EC4058BB  esmf_methodgridco        1287  ESMF_AttachMethods.F90
libesmf.so         000014A4ECAFE3FB  routine_run              3618  NUOPC_Driver.F90
libesmf.so         000014A4EBFCFB08  callVFuncPtr             2187  ESMCI_FTable.C
libesmf.so         000014A4EBFCF54C  ESMCI_FTableCallE         844  ESMCI_FTable.C
libesmf.so         000014A4EC30EC14  enter                    1216  ESMCI_VM.C
libesmf.so         000014A4EBFD01E0  c_esmc_ftablecall        1001  ESMCI_FTable.C
libesmf.so         000014A4EC505B28  esmf_compexecute         1273  ESMF_Comp.F90
libesmf.so         000014A4EC772344  esmf_gridcomprun         1927  ESMF_GridComp.F90
libesmf.so         000014A4ECAFF8B5  routine_executegr        3698  NUOPC_Driver.F90
libesmf.so         000014A4ECB014D8  executerunsequenc        3944  NUOPC_Driver.F90
libesmf.so         000014A4EC274E67  execute                   377  ESMCI_MethodTable.C
libesmf.so         000014A4EC274D05  c_esmc_methodtabl         317  ESMCI_MethodTable.C
libesmf.so         000014A4EC4058BB  esmf_methodgridco        1287  ESMF_AttachMethods.F90
libesmf.so         000014A4ECAFE3FB  routine_run              3618  NUOPC_Driver.F90
libesmf.so         000014A4EBFCFB08  callVFuncPtr             2187  ESMCI_FTable.C
libesmf.so         000014A4EBFCF54C  ESMCI_FTableCallE         844  ESMCI_FTable.C
libesmf.so         000014A4EC30EC14  enter                    1216  ESMCI_VM.C
libesmf.so         000014A4EBFD01E0  c_esmc_ftablecall        1001  ESMCI_FTable.C
libesmf.so         000014A4EC505B28  esmf_compexecute         1273  ESMF_Comp.F90
libesmf.so         000014A4EC772344  esmf_gridcomprun         1927  ESMF_GridComp.F90
access-om3-MOM6-C  000000000042620B  esmapp                    141  esmApp.F90
access-om3-MOM6-C  00000000004255DD  Unknown               Unknown  Unknown
libc-2.28.so       000014A4E828A865  __libc_start_main     Unknown  Unknown
access-om3-MOM6-C  00000000004254FE  Unknown               Unknown  Unknown

[manodeep]

Based on the traceback, this line seems to be the offending one, but the enclosing if condition should have prevented the divide-by-zero. Can sec_prod be negative?

[dougiesquire]

Can sec_prod be negative?

I don't think it's supposed to be, but I think it must be happening, possibly here.

[anton-seaice]

This was with 2025.07.000, its possible it's been fixed in newer versions, but it now fails with #41

[anton-seaice]

This still fails with MOM6@2026.01.000

How does

ticks = int(tick_prod) - (num_sec * ticks_per_second)

cause a

forrtl: error (73): floating divide by zero

Maybe I lack creativity, but thats a subtraction and a mutltiplication ?

[manodeep]

Hmm certainly not at first sight - may be see how ticks_per_second. And how did you figure out the line in the source?

[anton-seaice]

Oh hangon - I get the same failure through a different code path:

forrtl: error (73): floating divide by zero
Image              PC                Routine            Line        Source             
libpthread-2.28.s  000014B65508C990  Unknown               Unknown  Unknown
libimf.so          000014B6573FD247  __libm_log10_z0       Unknown  Unknown
access-om3-MOM6-C  0000000004305AF7  time_scalar_mult          778  time_manager.F90
access-om3-MOM6-C  00000000043062CA  time_scalar_divid         806  time_manager.F90
access-om3-MOM6-C  00000000043012DC  time_interp_list_         373  time_interp.inc
access-om3-MOM6-C  00000000042EEBE8  time_interp_exter         177  time_interp_external2.inc
access-om3-MOM6-C  0000000003E7530B  data_override_3d_        1638  data_override.inc
access-om3-MOM6-C  0000000003E6FC4F  data_override_2d_        1022  data_override.inc
access-om3-MOM6-C  0000000003E39201  ct_data_override_        3753  coupler_types.F90
access-om3-MOM6-C  0000000002293A27  ct_data_override          430  MOM_couplertype_infra.F90
access-om3-MOM6-C  0000000001079497  coupler_type_data         366  MOM_coupler_types.F90
access-om3-MOM6-C  0000000000548EAF  modeladvance             1957  mom_cap.F90
libesmf.so         000014B658A66227  execute                   377  ESMCI_MethodTable.C
libesmf.so         000014B658A660C5  c_esmc_methodtabl         317  ESMCI_MethodTable.C
libesmf.so         000014B658C19D4B  esmf_methodgridco        1287  ESMF_AttachMethods.F90
libesmf.so         000014B659364DFE  routine_run              2109  NUOPC_ModelBase.F90
libesmf.so         000014B6587BAA98  callVFuncPtr             2192  ESMCI_FTable.C
libesmf.so         000014B6587BA4DC  ESMCI_FTableCallE         849  ESMCI_FTable.C
libesmf.so         000014B658B31096  enter                    2730  ESMCI_VMKernel.C
libesmf.so         000014B658B1C7D4  enter                    1365  ESMCI_VM.C
libesmf.so         000014B6587BB170  c_esmc_ftablecall        1006  ESMCI_FTable.C
libesmf.so         000014B658D19FB8  esmf_compexecute         1274  ESMF_Comp.F90
libesmf.so         000014B658F89504  esmf_gridcomprun         1933  ESMF_GridComp.F90
libesmf.so         000014B65931E185  routine_executegr        3611  NUOPC_Driver.F90
libesmf.so         000014B65931FDA8  executerunsequenc        3857  NUOPC_Driver.F90
...

And how did you figure out the line in the source?

Just took a punt that it was correct for the source code version : https://github.com/NOAA-GFDL/FMS/tree/2025.03

ticks_per_second looks like it should be the default value (1)

[manodeep]

Ahhha log10 is involved 🙂