Feature/main ecbuild candidate 20220111 #29
Conversation
…solving conflicts
remove unused index bounds, and fix sum_h2 loop.
Update MOM_oda_incupd.F90
Fix to codecov file path search
Visc_rem_[uv] multiplied momentum budget diagnostics
Bugfix: a couple of mask_depth related issues
Make use_drag_rate independent of MEKE_damping
Merging dev/gfdl 08/05/2021
Fix to commit 893a06e which broke the SIGMA_SHELF_ZSTAR option
…0210727 IAU candidate 20210727
Fix register_diag_field axes_grp pointer init
This adds four new diagnostics building on the wind stress acceleration diagnostics du_dt_str, dv_dt_str (from mom-ocean#1437) - their thickness-weighted versions: h_du_dt_str, h_dv_dt_str (completing the set of diags from 3D thickness x momentum diagnostics mom-ocean#1398) - their viscous remnant fraction: du_dt_str_visc_rem, dv_dt_str_visc_rem (completing the set of diags from Visc_rem_[uv] multiplied momentum budget diagnostics ocean-eddy-cpt#10) Nora did some quick tests with the CPT NeverWorld2 setup, which confirm that online and offline multiplication by 1) h or 2) visc_rem_[uv] coincide (up to 1) interpolation error and 2) numerical noise, respectively), and illustrated this beautifully with some figures that accompanied the first of the three commits that were squashed into one.
Cleaned up 26 falsely optional or unused arguments in the vertical diffusivity code, and related changes. Several descriptive comments were also corrected, including the correction of the units of 10 variables related to CVMix_KPP. This commit includes: - Made the Kd_int arguments to set_diffusivity() and 3 subsidiary routines mandatory and reordered the arguments so that the non-optional arguments come before the grid types - Made the halo_TS and double_diffuse arguments to set_diffusivity_init() mandatory. - Made the Time argument to ALE_sponge() mandatory. - Made the Kd and Kv arguments to calculate_CVMIX_conv() mandatory. - Removed the unused halo argument to adjust_salt(). - Removed the unused Kddt_convect argument to full_convection(). - Made the halo arguments to full_convection()and smoothed_dRdT_dRdS() mandatory. - Made the useALEalgorithm argument to geothermal_init() mandatory. - Removed the unused initialize_all arguments to Calculate_kappa_shear() and Calc_kappa_shear_vertex(). - Removed the unused I_Ld2_1d and dz_Int_1d arguments to kappa_shear_column(). - Made 3 arguments to calculate_projected_state() mandatory and reordered the arguments accordingly. - Eliminating the unused skip_diags arguments to calculateBuoyancyFlux() and extractFluxes(), which are now effectively always false. All answers are bitwise identical, and no output is changed.
Moved the routine rotate_dyngrid() from the MOM_transcribe_grid module to rotate_dyn_horgrid() in the MOM_dyn_horgrid module so that this routine can also be used at some point by SIS2 to implement rotational consistency testing, and also to reflect that this routine only works with types from its new module. The two routines are the same apart from some added comments, and the old name of rotate_dyngrid() is still available from MOM_transcribe_grid via a module use statement. All answers are bitwise identical.
Minimized the dependence on dyn_horgrid in initialize_MOM by working directly with the horizontal index type whereever possible and by moving the calls that create the MOM_grid_type earlier in the routine, to limit the duration of the dyn_horgrid_type, and to better co-locate grid-related parameters in the parameter_doc files. Also uses the new interface to rotate_dyn_horgrid from the MOM_dyn_horgrid module in place of the rotate_dyngrid interface from the MOM_transcribe_grid module. All answers are bitwise identical, but the order of some entries in the MOM_parameter_doc files has changed.
Modified drivers/unit_drivers/MOM_sum_driver.F90 to compile with the latest version of the rest of the MOM6 code by using the proper types in the various initialization calls, and verified that it runs as intended.
guillaumevernieres
requested review from
travissluka,
liuxiao37k,
kbhargava,
clouden90 and
HamidehGMAO
|
my build on discover with intel fails with this error
|
|
@HamidehGMAO , you need to edit the soca/bundle/CMakeLists.txt and replance the mom6 branch with this one. |
@guillaumevernieres just did it, but got the same error |
|
I was able to successfully build soca bundle on both Orion and Hera (Intel compilers) with BGC option |
|
I have the same issue on Orion as @liuxiao37k . @guillaumevernieres My log files are at |
|
FYI - I also tested with soca-science; build was successful, but cycling ctests failed due to same segmentation error during |
|
Thanks for testing @liuxiao37k, @HamidehGMAO and @kbhargava , I'll have a look. What intel version are we using on Hera @liuxiao37k ? |
@guillaumevernieres: module load jedi/intel-impi/2020.2 |
|
I don't know ... It seems to work for me now (it was failing earlier today):
4, 45 and 52 fail in the compare step but the other tests are OK. |
There was a problem hiding this comment.
whew! works for me now. All soca ctests passed except the three that you @guillaumevernieres identified. I also ran soca-science ctests including all BGC ones. No issues. So I am approving the PRs before something goes wrong again...
…args +(*)Make unit_scale_type arguments non-optional
Description
closes #27