Fixing openmp compile

[nikizadehgfdl]

• openmp threads produce non-openmp stats
• openmp 1 thread run has to be a test. Otherwise it would be safer not supporting openmp at all.

[codecov-io]

Codecov Report

Merging #1010 into dev/gfdl will not change coverage.
The diff coverage is n/a.

@@            Coverage Diff            @@
##           dev/gfdl    #1010   +/-   ##
=========================================
  Coverage     43.16%   43.16%           
=========================================
  Files           213      213           
  Lines         62352    62352           
=========================================
  Hits          26915    26915           
  Misses        35437    35437

Impacted Files	Coverage Δ
...rc/parameterizations/vertical/MOM_diabatic_aux.F90	62.13% <ø> (ø)	⬆️
src/diagnostics/MOM_diagnostics.F90	87.38% <ø> (ø)	⬆️
src/tracer/MOM_tracer_advect.F90	69.03% <ø> (ø)	⬆️
...ameterizations/lateral/MOM_mixed_layer_restrat.F90	86.31% <ø> (ø)	⬆️
src/parameterizations/lateral/MOM_hor_visc.F90	62.67% <ø> (ø)	⬆️
src/core/MOM_barotropic.F90	71.92% <ø> (ø)	⬆️
src/core/MOM_PressureForce_blocked_AFV.F90	0% <ø> (ø)	⬆️
...arameterizations/lateral/MOM_thickness_diffuse.F90	63.7% <ø> (ø)	⬆️
...c/parameterizations/vertical/MOM_vert_friction.F90	72.12% <ø> (ø)	⬆️
...eterizations/lateral/MOM_lateral_mixing_coeffs.F90	59.61% <ø> (ø)	⬆️
... and 2 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9ca5fb9...729a497. Read the comment docs.

[nikizadehgfdl]

This fix is not complete. Dev/gfdl still has openmp issues. I don't know how I was able to compile this branch!!

[nikizadehgfdl]

This branch fixes -openmp compilation issues as well as answer changes for threaded runs.
It closes issue #973
OMP threaded runs (ocean_nthreads>1) still underperform.

[nikizadehgfdl]

This fixes the answers mismatch between openmp and non-openmp execs.

[marshallward]

I just tested an OpenMP build and am also seeing errors in the MOM_geothermal.F90 code:

ftn -Duse_libMPI -Duse_netCDF -DSPMD -Duse_LARGEFILE  -I/opt/cray/pe/netcdf/4.4.0/GNU/5.1/include -fcray-pointer -fdefault-real-8 -fdefault-double-8 -Waliasing -ffree-line-length-none -fno-range-check -O2 -fno-expensive-optimizations -fopenmp -I /lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../deps/fms/build  -c -I/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../config_src/dynamic_symmetric -I/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/framework	/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/parameterizations/vertical/MOM_geothermal.F90
/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/parameterizations/vertical/MOM_geothermal.F90:158:60:

                   .or. CS%id_internal_heat_temp_tendency > 0
                                                            1
Error: Unexpected assignment statement at (1)
/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/parameterizations/vertical/MOM_geothermal.F90:161:60:

                   .or. CS%id_internal_heat_temp_tendency > 0
                                                            1
Error: Unexpected assignment statement at (1)
/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/parameterizations/vertical/MOM_geothermal.F90:163:65:

   if (CS%id_internal_heat_heat_tendency > 0) work_3d(:,:,:) = 0.0
                                                                 1
Error: Unexpected simple IF statement at (1)
/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/parameterizations/vertical/MOM_geothermal.F90:164:39:

   if (compute_h_old) h_old(:,:,:) = 0.0
                                       1
Error: Unexpected simple IF statement at (1)
/lustre/f2/dev/gfdl/Marshall.Ward/dev/mom6/.testing//../src/parameterizations/vertical/MOM_geothermal.F90:165:39:

   if (compute_T_old) T_old(:,:,:) = 0.0
                                       1
Error: Unexpected simple IF statement at (1)

Assuming it's not too much trouble, I think we should probably add this to the test suite, if only to check that it builds.

Edit: Niki's patch will fix this issue (along with others)

[marshallward]

Gaea regression tests: https://gitlab.gfdl.noaa.gov/ogrp/MOM6/pipelines/9074

[marshallward]

Gaea regressions have passed.

I can also build this internally although I am seeing some differences in answers for some of the TCs in the new test suite. But I will look into those in a different PR.