Sync with MPAS-Dev v8.4.2 release - #287
Open
clark-evans wants to merge 17 commits into
Open
Conversation
…PRINT macro Registry-generated code in setup_packages.inc (generated by the gen_pkg_debug_info registry function in gen_inc.c) may contain continuation lines in the invocation of the PACKAGE_LOGIC_PRINT macro, and this can lead to compilation failures with the nvfortran compiler. To work around this issue, this commit avoids continuation lines with the PACKAGE_LOGIC_PRINT macro only when compiling with the nvfortran or pgf90 compilers. This may result in source lines longer than 132 columns, but in practice, recent releases of the nvfortran compiler appear to have no problems with any of the source lines generated in the current version of MPAS.
…om 2 to 4 This commit changes the default value of the atmosphere core namelist option config_number_of_sub_steps from 2 to 4 in the Registry.xml file. This change addresses an issue in convection-permitting simulations where noise can sometimes appear in the jet streams using the previous default value (config_number_of_sub_steps = 2). The noise, when present, also produces unphysical features in horizontal kinetic energy spectra at high wavenumbers.
…4.2 (PR MPAS-Dev#1495) This merge modifies the Registry code generation to work around an issue with continuation lines in instances of the PACKAGE_LOGIC_PRINT macro when building with the nvfortran compiler. Registry-generated code in setup_packages.inc (generated by the gen_pkg_debug_info registry function in gen_inc.c) may contain continuation lines in the invocation of the PACKAGE_LOGIC_PRINT macro, and this can lead to compilation failures with the nvfortran compiler. To work around this issue, this merge avoids continuation lines with the PACKAGE_LOGIC_PRINT macro only when building with the nvfortran or pgf90 compilers (or, more generally, when the CPRPGI macro is defined). This may result in source lines longer than 132 columns, but in practice, recent releases of the nvfortran compiler appear to have no problems with any of the source lines generated in the current version of MPAS. * framework/workaround_nhvpc_macro_issue: Work around nvfortran issue with continuation lines in PACKAGE_LOGIC_PRINT macro
…PAS-Dev#1496) This merge changes the default value of the atmosphere core namelist option config_number_of_sub_steps from 2 to 4 in the Registry.xml file. This change addresses an issue in convection-permitting simulations where noise can sometimes appear in the jet streams using the previous default value (config_number_of_sub_steps = 2). The noise, when present, also produces unphysical features in horizontal kinetic energy spectra at high wavenumbers. Note: Increasing the number of sub-steps from 2 to 4 can be expected to increase the computational cost of the dynamics by ~10%, and the computational cost of a full-physics simulation by ~5%. * atmosphere/sub_steps_default_4: Change default value of atmosphere core config_number_of_sub_steps from 2 to 4
…xml file
This commit swaps the order of 'lbc_ni' and 'lbc_nr' in the atmosphere core's
Registry.xml file so that the LBC fields for the 'ni' and 'nr' scalars are
defined in the same order as the tendencies ('tend_ni' and 'tend_nr') and state
('ni' and 'nr') for these scalars.
Note that unless LBCs are provided for 'ni' and 'nr', the previously incorrect
order of 'lbc_ni' and 'lbc_nr' would not affect results, as both fields would
be filled with zero values.
…PAS-Dev#1498) This merge swaps the order of 'lbc_ni' and 'lbc_nr' in the atmosphere core's Registry.xml file so that these LBC fields for the 'ni' and 'nr' scalars are defined in the same order as the tendencies ('tend_ni' and 'tend_nr') and state ('ni' and 'nr') for these scalars. Note that unless LBCs are provided for 'ni' and 'nr', the previously incorrect order of 'lbc_ni' and 'lbc_nr' would not affect results, as both fields would be filled with zero values. Thanks to Trude Eidhammer (NSF NCAR / RAL; @trudeeidhammer) for identifying the issue addressed in this merge. * atmosphere/fix_lbc_ni_nr_order: Correct order of lbc_ni and lbc_nr in the atmosphere core's Registry.xml file
This commit fixes the OpenACC data movement for the field rho_zz in routine mpas_atm_pre_compute_solve_diagnostics. Previously, an OpenACC create statement was used for this field. This is still correct for the cold initialization case , but produces incorrect results in the case of a restarted run. This has now been changed to a copyin statement. This issue was discovered as a result of restarted real simulations on GPUs crashing. The unphysical values from the dynamical core lead to a segmentation fault occuring in the SW radiation calls. Furthermore, the Jablonowski- Williamson baroclinic wave case also showed discrepancies between the cold initialization and the restarted test case. Both of these issues have been addressed by this commit.
The library built from src/external/esmf_time_f90 contains MPAS's bundled
ESMF time-manager stubs, not the full ESMF. Naming the output file libesmf
lets the dynamic loader substitute a real ESMF library found via
LD_LIBRARY_PATH, which breaks every MPAS executable.
Set the target's OUTPUT_NAME to esmf_time so the library is built and
installed as libesmf_time.{so,a}, resolving the name conflict. The CMake
target and alias names are unchanged.
This merge addresses a lack of TKE in MPAS-A with the 1.5-order prognostic TKE option when there is no TKE in the initial conditions (for example, in real-data cases). In the prognostic TKE option, the source terms depend on non-zero TKE through the horizontal and vertical diffusion coefficients, and so if the ICs contain no TKE, no TKE will be generated. This merge addresses this issue by setting a small lower-bound (0.01 m^2/s) on the 'eddy_visc_horz' and 'eddy_visc_vert' fields in the les_models() routine. * lesfix-8.4.1: add minimum value for Kh and Kv to allow for zero initial tke
…MPAS-Dev#1503) This merge corrects the OpenACC data movement for the field rho_zz in routine mpas_atm_pre_compute_solve_diagnostics. Previously, an OpenACC 'create' statement was used for this field. This was correct for cold-start simulations, but produces incorrect results in the case of a restarted simulation. This merge changes the 'create' statement to a 'copyin' statement, thereby correcting the model restart capability when running on GPUs. This issue was discovered as a result of restarted real-data simulation on GPUs crashing. The unphysical values from the dynamical core lead to a segmentation fault occurring in the SW radiation calls. Furthermore, the Jablonowski and Williamson baroclinic wave case also showed discrepancies between the cold initialization and the restarted test case. Both of these issues have been addressed by this merge. * atmosphere/fix_acc_data_movement: Fix for GPU-enabled restart runs
…ev#1506) This merge renames the internal ESMF time manager library to libesmf_time in CMake builds of MPAS, matching the name used in GNU Make builds. The library built from src/external/esmf_time_f90 contains MPAS's bundled ESMF time-manager, not the full ESMF. Prior to this merge, naming the library file libesmf allowed the dynamic loader substitute a full ESMF library found via LD_LIBRARY_PATH, breaking the MPAS executable. By setting the target's OUTPUT_NAME to esmf_time in src/external/esmf_time_f90/CMakeLists.txt so the library is built and installed as libesmf_time.{so,a}, this issue can be avoided. Note that the CMake target and alias names are unchanged. * bugfix/external-esmf-fix: Rename bundled ESMF time-manager library output to esmf_time
This merge addresses several issues in the MPAS-Atmosphere model and in the MPAS infrastructure. Specific changes include: * Fix an issue with the naming of MPAS's internal ESMF time-keeping library in CMake builds of MPAS. (PR MPAS-Dev#1506) * Fix an issue with restart runs in MPAS-Atmosphere on GPUs. (PR MPAS-Dev#1503) * Address a lack of TKE in MPAS-Atmosphere with the 1.5-order prognostic TKE option when there is no TKE in the initial conditions (for example, in real-data cases). (PR MPAS-Dev#1499) * Correct the ordering of the 'lbc_ni' and 'lbc_nr' scalars in the atmosphere core's Registry.xml file. Note that unless LBCs are provided for 'ni' and 'nr', the previously incorrect order of 'lbc_ni' and 'lbc_nr' would not affect results, as both fields would be filled with zero values. (PR MPAS-Dev#1498) * Change the default value of the config_number_of_sub_steps namelist option from 2 to 4 in the atmosphere core's Registry.xml file to address an issue in convection-permitting simulations where noise can sometimes appear in the jet streams using the previous default value of 2. The noise, when present, also produces unphysical features in horizontal kinetic energy spectra at high wavenumbers. (PR MPAS-Dev#1496) Note: Increasing the number of sub-steps from 2 to 4 can be expected to increase the computational cost of the dynamics by ~10% and the computational cost of a full-physics simulation by ~5%. * Work around an issue with continuation lines in instances of the PACKAGE_LOGIC_PRINT macro that prevented successful compilation of MPAS-Atmosphere with the 'nvfortran' compiler. (PR MPAS-Dev#1495) * hotfix-v8.4.2: Rename bundled ESMF time-manager library output to esmf_time Fix for GPU-enabled restart runs Correct order of lbc_ni and lbc_nr in the atmosphere core's Registry.xml file add minimum value for Kh and Kv to allow for zero initial tke Change default value of atmosphere core config_number_of_sub_steps from 2 to 4 Work around nvfortran issue with continuation lines in PACKAGE_LOGIC_PRINT macro Update version number to 8.4.2
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR brings in the bugfixes from the MPAS-Dev v8.4.2 release. The only merge conflicts were:
The bugfixes generally do not impact our forecasts, so I expect this to not be answer changing.
This should not be squash-merged to ensure the commit history from upstream is retained.
Mandatory Questions
Reviews