Some ocean_solo tests crash when compiled with -openmp

[nikizadehgfdl]

The following ocean_solo test cases crashing in repro-openmp mode (but pass in repro mode)
mom6_solo_DOME, mom6_solo_nonBous_global, mom6_solo_Phillips_2layer

The following test case runs but does not reproduce non-openmp timestats:
mom6_solo_single_column_z_BATS

[nikizadehgfdl]

Here are the tracebacks for these crashes:

DOME and Phillips_2layer :

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
fms_mom6_solo_com  00000000006CEF69  mom_thickness_dif         203  MOM_thickness_diffuse.F90
fms_mom6_solo_com  0000000000465820  mom_mp_step_mom_          999  MOM.F90
fms_mom6_solo_com  0000000000401D0B  MAIN__                    396  MOM_driver.F90

nonBous_global:

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
fms_mom6_solo_com  0000000000675D41  mom_eos_wright_mp         353  MOM_EOS_Wright.F90
fms_mom6_solo_com  000000000062413D  mom_eos_mp_int_sp         402  MOM_EOS.F90
fms_mom6_solo_com  0000000000763561  mom_initializatio        1627  MOM_initialization.F90
fms_mom6_solo_com  0000000000771D50  mom_initializatio         506  MOM_initialization.F90
fms_mom6_solo_com  000000000047D370  mom_mp_initialize        1754  MOM.F90
fms_mom6_solo_com  0000000000400C80  MAIN__                    251  MOM_driver.F90

[Zhi-Liang]

Hi Niki,

I will take a look to fix the issue.

Thanks for the information!

Zhi

On Tue, Mar 18, 2014 at 3:50 PM, nikizadehgfdl [email protected]:

The following ocean_solo test cases crashing in repro-openmp mode (but
pass in repro mode)
mom6_solo_DOME, mom6_solo_nonBous_global, mom6_solo_Phillips_2layer,
mom6_solo_circle_obcs

The following test case runs but does not reproduce non-openmp timestats:
mom6_solo_single_column_z_BATS

Reply to this email directly or view it on GitHubhttps://github.com//issues/7
.

[Zhi-Liang]

Hi Niki,

I just push some changes in the Github and submit a pull request. I
tested with these changes in experiment
global, DOME, nonBous_global and single_column_z. When compiling with
-openmp and REPRO mode,
All the experiments runs OK and reproduce answer when compiling without
-openmp. Also global, DOME, nonBous_global reproduce between 1 threads and
4 threads. If compiling with production mode (-O3), there is no guarantee
that it reproduce between compiling with openmp and without openmp.

Could you test it out?

Thanks,

Zhi

On Tue, Mar 18, 2014 at 4:17 PM, nikizadehgfdl [email protected]:

Here are the tracebacks for these crashes:

DOME and Phillips_2layer :
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
fms_mom6_solo_com 00000000006CEF69 mom_thickness_dif 203
MOM_thickness_diffuse.F90
fms_mom6_solo_com 0000000000465820 mom_mp_step_mom_ 999 MOM.F90
fms_mom6_solo_com 0000000000401D0B MAIN__ 396 MOM_driver.F90

nonBous_global:

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
fms_mom6_solo_com 0000000000675D41 mom_eos_wright_mp 353 MOM_EOS_Wright.F90
fms_mom6_solo_com 000000000062413D mom_eos_mp_int_sp 402 MOM_EOS.F90
fms_mom6_solo_com 0000000000763561 mom_initializatio 1627
MOM_initialization.F90
fms_mom6_solo_com 0000000000771D50 mom_initializatio 506
MOM_initialization.F90
fms_mom6_solo_com 000000000047D370 mom_mp_initialize 1754 MOM.F90
fms_mom6_solo_com 0000000000400C80 MAIN__ 251 MOM_driver.F90

Reply to this email directly or view it on GitHubhttps://github.com//issues/7#issuecomment-37982294
.

[nikizadehgfdl]

This is issue is fixed. All RTS compile and run with -openmp and reproduce the non-openmp timestats (when using -d 1 for ocean).