oceanG works, but oceanM fails, version 631/629

[stone]

Hi all,
I come across a strange error when building test case "upwelling". I am updating to ROMS3.6 (try version 631/629). When I turn DEBUG on, it works well. However, when I urn DEBUG off, the run-time error "MPI: On host, Program oceanM, Rank 0, Process 21092 received signal SIGSEGV(11)" came after outputting

Code: Select all

NL ROMS/TOMS: started time-stepping: (Grid: 01 TimeSteps: 00000001 - 00001440)

. It stops before this as it should in DEBUG mode

Code: Select all

   STEP   Day HH:MM:SS  KINETIC_ENRG   POTEN_ENRG    TOTAL_ENRG    NET_VOLUME
          C => (i,j,k)       Cu            Cv            Cw         Max Speed

      0     0 00:00:00  0.000000E+00  6.585677E+02  6.585677E+02  3.884376E+11
            (00,00,00)  0.000000E+00  0.000000E+00  0.000000E+00  0.000000E+00
      DEF_HIS   - creating history file: ocean_his.nc
      WRT_HIS   - wrote history  fields (Index=1,1) into time record = 0000001
      DEF_AVG   - creating average file: ocean_avg.nc
      DEF_DIAGS - creating diagnostics file: ocean_dia.nc
      1     0 00:05:00  6.479408E-13  6.585677E+02  6.585677E+02  3.884376E+11
            (01,01,01)  1.069435E-09  1.950204E-06  0.000000E+00  9.216783E-06

I am using ifort 10, with SGI built-in libmpi, netcdf3.6; It works for OLD versions and the DEBUG mode of this version; Is there possible a bug?

[kate]

There is a bug, but I don't know that its in ROMS. Does the serial code work for you? Does it run with say -O2 instead of -O3 (or whatever you are using). Can you try other compilers or versions of ifort?

[stone]

Sorry, I forgot to mention that oceanS does not work either, and ifort -O2/-O1 both failed.
I am not the SGI administrator, I will try to contact the administrator to install some new version of ifort. But I don't understand why oceanG works for the same configuration except some compiling options(-g -check uninit -ftrapuv -traceback).