| f6a6685 | Luca Bertagna | 10 February 2022, 23:34:11 UTC | HOMME: pre-processor mods to handle SCREAM macro In a few places, we trigger the same behavior as for the CAM macro. | 11 February 2022, 00:03:09 UTC |
| ba3528c | noel | 09 February 2022, 22:28:50 UTC | Merge branch E3SM-Project/ndk/machinefiles/PM-NODENAME_REGEX (PR #4780) Correct NODENAME_REGEX for Perlmutter | 09 February 2022, 22:28:50 UTC |
| fba8cd0 | Jon Wolfe | 09 February 2022, 19:45:18 UTC | Merge branch 'akturner/seaice/single-cell-test-case' (PR #4620) Add single cell test case Makes the following changes to mpassi standalone testing, which is not used by E3SM: * Updates single cell namelist file * Adds single cell test case with simple plotting of results * Updates data download script Fixes #4619 [BFB] | 09 February 2022, 19:45:18 UTC |
| af24574 | Wuyin Lin | 09 February 2022, 18:25:22 UTC | Merge branch 'wlin/atm/fix_narrm_4amip' (PR #4783) Enable northamericax4v1pg2_WC14to60E2r3 for AMIP DOCN added for searching component pattern in compset for the North America RRM. This is to enable the grid for F-case simulations which uses DOCN. A 30-node pe-layout (M) and a 60-node pe-layout (L) are also added for running on chrysalis with this grid. Fixed issue #4745 [BFB] | 09 February 2022, 18:25:22 UTC |
| 36bf9a9 | Wuyin Lin | 08 February 2022, 21:14:00 UTC | Enable northamericax4v1pg2_WC14to60E2r3 for AMIP DOCN added for searching component pattern in compset for the North America RRM. This is to enable the grid for F-case simulations which uses DOCN. A 30-node pe-layout (M) and a 60-node pe-layout (L) are also added for running on chrysalis with this grid. [BFB] | 08 February 2022, 21:14:00 UTC |
| 447d2ba | Jon Wolfe | 08 February 2022, 16:20:56 UTC | Merge branch 'trhille/framework/add-mostRecentAccessTime-attribute-to-streams' (PR #4752) Add mostRecentAccessTime attribute to streams Add a new attribute to streams called 'mostRecentAccessTime', which is useful when multiple time-levels are needed at once from an input file. The actualWhen argument to the MPAS_stream_mgr_read() routine is replaced with an optional argument saveActualWhen. The actualWhen argument was ambiguous when MPAS_stream_mgr_read was called in a way such that multiple streams were read; the actualWhen value returned came from the final stream read and could possibly be wrong for the other streams that were read. This also changes the land ice core to stop using actualWhen and replace it with saveActualWhen. Neither the ocean nor seaice cores use actualWhen so no other changes are necessary. [BFB] | 08 February 2022, 16:20:56 UTC |
| d53434b | noel | 07 February 2022, 18:32:09 UTC | Correct nodename regex for PM | 07 February 2022, 18:32:09 UTC |
| 44814ae | Azamat Mametjanov | 03 February 2022, 21:50:52 UTC | Merge branch 'azamat/jlse/update-oneapi-module' (PR #4775) Update JLSE OneAPI module. [BFB] | 03 February 2022, 21:50:52 UTC |
| 5aeac51 | Azamat Mametjanov | 03 February 2022, 21:39:42 UTC | Update JLSE OneAPI module | 03 February 2022, 21:39:42 UTC |
| 740c4b2 | Jon Wolfe | 03 February 2022, 18:09:53 UTC | Merge branch 'philipwjones/mpas-framework/reprosums' (PR #4700) Add new reproducible global sum module for MPAS components Adds new module to compute global sums of distributed arrays in a reproducible way so that they are independent of summation order (eg from different partitions). The 64-bit real algorithm is the same as the DDPDD in shr_reprosum but this interface supports some features that were added for internal MPAS use. [BFB] | 03 February 2022, 18:09:53 UTC |
| ff1f175 | noel | 03 February 2022, 16:37:18 UTC | Merge branch E3SM-Project/ndk/machinefiles/perlmutter-fix-NODENAME_REGEX (PR #4773) For Perlmutter, change NODENAME_REGEX to be "perlmutter" | 03 February 2022, 16:37:18 UTC |
| acc4d50 | Matthew Hoffman | 03 February 2022, 02:00:12 UTC | Only modify the mostRecentAccessTime is a stream was actually read Before this commit, a default value of '0000-01-01_00:00:00' was being saved to the new mostRecentAccessTime attribute on calls to mpas_stream_mgr_read() where a stream wasn't actually read because its alarm wasn't ringing. This commit initializes actualWhen_local to a non-time value in order to only set mostRecentAccessTime if actualWhen_local is successfully updated to a time by read_stream(). | 03 February 2022, 02:00:12 UTC |
| fbf5530 | noel | 02 February 2022, 20:52:23 UTC | Change the NODENAME_REGEX to be perlmutter to avoid conflicts with machines using "login". This change was possible after https://github.com/ESMCI/cime/issues/4133 It does require that users have proper setting of env var NERSC_HOST. [bfb] | 02 February 2022, 20:52:23 UTC |
| bea0ad7 | noel | 02 February 2022, 20:04:41 UTC | Merge branch E3SM-Project/ndk/machinefiles/gcp-pelayout-adjust-for-tests (PR #4766) For GCP, adjust default layouts for some common tests | 02 February 2022, 20:04:41 UTC |
| f33f684 | Azamat Mametjanov | 02 February 2022, 19:51:56 UTC | Merge branch 'azamat/bebop/set-pioroot1-erp' (PR #4772) Run low-res ERP test with pio_root=1 to fix a reported memleak. [BFB] | 02 February 2022, 19:51:56 UTC |
| d9bbcc4 | Matthew Hoffman | 02 February 2022, 18:31:41 UTC | Update comment description of MPAS_stream_mgr_read() Updated comments explain why the saveActualWhen argument is needed and why it is not possible to just always save the actualWhen time. | 02 February 2022, 18:32:21 UTC |
| c8ca9fc | Matthew Hoffman | 02 February 2022, 18:25:26 UTC | Revert "Remove saveActualWhen argument to mpas_stream_mgr_read()" This reverts commit 1baed5ebfaaa36165d1eb677f8a62806b97b2f4a. This revert goes with the revert in the previous commit. See commit message there for details. | 02 February 2022, 18:25:26 UTC |
| f822d68 | Matthew Hoffman | 02 February 2022, 18:22:31 UTC | Revert "Use a dummy value for actualWhen if a stream does not include xtime" This reverts commit c7817c6370a5d56d86a78cab543e3b09a5c8de29. The attempt to parse i/o error codes was not reliable because tests always returned MPAS_IO_ERR_PIO instead of more specific, useful error codes. Thus it was not possible to differentiate between xtime being missing from a file and any other i/o error that should be treated as a real error. | 02 February 2022, 18:22:31 UTC |
| 5834afa | Peter Schwartz | 02 February 2022, 17:01:42 UTC | Merge branch 'evasinha/lnd/mod_param_harvest' (PR #4757) Modified parameters for miscanthus and switchgrass based on calibration and validation finalized for revised manuscript submitted in GMD https://gmd.copernicus.org/preprints/gmd-2021-244/ Minor edits in estimation of perennial crop harvest. [nonBFB] for crop tests. Fixes #4608 | 02 February 2022, 17:01:42 UTC |
| c7817c6 | Matthew Hoffman | 02 February 2022, 16:49:37 UTC | Use a dummy value for actualWhen if a stream does not include xtime Some streams do not have the xtime variable. In those cases, attempting to access xtime leads to an error. This commit uses a dummy value for actualWhen if xtime is not present in an input file. | 02 February 2022, 16:49:37 UTC |
| 9f3096d | Jon Wolfe | 02 February 2022, 15:51:56 UTC | Merge branch 'jonbob/scripts/add-jra-orrs18to6v3' (PR #4756) Add support for the oRRS18to6v3 grid to run with JRA forcing Adds a grid resolution and all necessary mapping files to support the high-res ocn/ice grid oRRS18to6v3 run with JRA forcing, which is on a TL319 datm grid. [BFB] | 02 February 2022, 15:51:56 UTC |
| afbf194 | Jon Wolfe | 02 February 2022, 15:48:13 UTC | Merge branch 'mark-petersen/ocn/init-mode-streams' (PR #4769) Add mode specification to conservation check streams Currently, the conservation check analysis member does not specify its MPAS mode. This was an oversight. Without including this, the conservation check stream is included in MPAS init mode. This can cause an error, and it is never needed for init mode anyway. All the other analysis members include these lines. [BFB] | 02 February 2022, 15:48:13 UTC |
| 1baed5e | Matthew Hoffman | 02 February 2022, 02:21:00 UTC | Remove saveActualWhen argument to mpas_stream_mgr_read() Having saving of the actualWhen time being optional could lead to confusion if one accesses the new mostRecentAccessTime attribute elsewhere. This commit makes it so the actualWhen time is always saved for all streams that are read. Hence the saveActualWhen argument is deleted. | 02 February 2022, 02:21:00 UTC |
| 767728d | Phil Jones | 02 February 2022, 01:21:47 UTC | added flag for IEEE compliance for reproducible sums some compiler options interfere with the reproducible global sum algorithm, so this flag prevents these optimization for the global sum module | 02 February 2022, 01:21:47 UTC |
| d70ba2e | Jon Wolfe | 01 February 2022, 22:09:54 UTC | Merge branch 'xylar/mpas-o/fix-init-realistic-coriolis' (PR #4750) Initialize Coriolis param in its own subroutine In situations where we do not expand the Earth sphere radius, the Coriolist parameter still needs to get initialized to a realistic value. This merge creates a new subroutine for initializing the Coriolis parameter that is called regardless of whether the Earth sphere is being expanded or not. This bug was reported in: MPAS-Dev/compass#304 This only affects MPAS-Ocean in init mode and is bit-for-bit in E3SM. [BFB] | 01 February 2022, 22:09:54 UTC |
| 098ae6c | noel | 01 February 2022, 16:58:35 UTC | Merge branch E3SM-Project/ndk/machinefiles/PM-default-layouts (PR #4771) Change default 1-node layouts for perlmutter to be CPU/GPU aware | 01 February 2022, 16:58:35 UTC |
| b71fa6e | Azamat Mametjanov | 01 February 2022, 01:17:44 UTC | Merge branch 'azamat/tests/add-gpu-tests' (PR #4762) Add OpenACC, OpenMP and CXX GPU tests [BFB] | 01 February 2022, 01:17:44 UTC |
| 49f3c27 | Azamat Mametjanov | 01 February 2022, 00:47:46 UTC | Run low-res ERP test with pio_root=1 Also add a python3 module on Bebop | 01 February 2022, 00:47:46 UTC |
| bc7b877 | Azamat Mametjanov | 01 February 2022, 00:40:49 UTC | Cleanup pelayouts for Blues | 01 February 2022, 00:40:49 UTC |
| 9f9c196 | Azamat Mametjanov | 01 February 2022, 00:38:45 UTC | Cleanup pelayouts for Blues | 01 February 2022, 00:38:45 UTC |
| 8b2f4c5 | The Excavator | 31 January 2022, 20:13:18 UTC | Make sure to use HDF5_USE_FILE_LOCKING=FALSE even when not using threads. | 31 January 2022, 20:13:18 UTC |
| 1b9422b | noel | 31 January 2022, 17:08:36 UTC | Change default layout setting for perlmutter to use 64 MPI's when using cpu and 4 MPI's when gpu | 31 January 2022, 17:08:36 UTC |
| acdc603 | Phil Jones | 31 January 2022, 17:01:11 UTC | changed filename for Cmake build | 31 January 2022, 17:01:11 UTC |
| 9ba796b | Mark Petersen | 30 January 2022, 05:23:17 UTC | Add mode=forward;analysis to conservation check | 30 January 2022, 05:25:40 UTC |
| 932c5b2 | Azamat Mametjanov | 29 January 2022, 02:21:49 UTC | Add pgi version check | 29 January 2022, 02:21:49 UTC |
| 464f8cd | Benjamin Hillman | 28 January 2022, 22:43:31 UTC | Merge branch 'whannah/mmf/remove-test-tasks' (#4760) The MMFXX test with ESMT enabled was still using a specific task/thread count even though we had fixed the issue that had forced us to use that approach, so this PR remedies this. [BFB] but requires new baseline for changed test name | 28 January 2022, 22:43:31 UTC |
| ffaaf4a | Phil Jones | 28 January 2022, 22:29:30 UTC | renamed MPAS global sum routines to match underscore convention | 28 January 2022, 22:29:30 UTC |
| f5a2037 | Phil Jones | 28 January 2022, 22:24:28 UTC | fixed missing explicit R4 declarations in MPAS global sum module | 28 January 2022, 22:24:28 UTC |
| b085dc0 | Azamat Mametjanov | 28 January 2022, 00:12:35 UTC | Add OpenACC, OpenMP and CXX GPU tests | 28 January 2022, 00:12:35 UTC |
| 78a51e7 | Walter Hannah | 27 January 2022, 21:26:51 UTC | remove task spec for MMF test w/ ESMT | 27 January 2022, 23:16:44 UTC |
| 32ce6e8 | James Foucar | 27 January 2022, 22:42:37 UTC | Merge branch 'bartgol/homme/remove-scream-ifndef-occurrence' into master (PR #4761) HOMME: remove one occurrence of the CPP ifndef SCREAM The guarded function (push_forcing_to_c) is used unguarded later on, so a link error would ensue. Since the C function is compiled regardless of the SCREAM macro, the simple solution is to remove this ifndef guard. NOTE: this should NOT affect any code/build that does not define the SCREAM macro. Such codes/builds were already preprocessing the file as if the ifndef wasn't there at all. [BFB] * bartgol/homme/remove-scream-ifndef-occurrence: HOMME: remove one occurrence of the CPP ifndef SCREAM | 27 January 2022, 22:42:37 UTC |
| 0590e44 | Luca Bertagna | 27 January 2022, 22:33:40 UTC | HOMME: remove one occurrence of the CPP ifndef SCREAM The guarded function (push_forcing_to_c) is used unguarded later on, so a link error would ensue. Since the C function is compiled regardless of the SCREAM macro, the simple solution is to remove this ifndef guard. NOTE: this should NOT affect any code/build that does not define the SCREAM macro. | 27 January 2022, 22:33:40 UTC |
| 5f05bf9 | Peter Schwartz | 27 January 2022, 20:14:24 UTC | Merge branch 'azamat/elm/fix-datatype-memleaks' (PR #4711) Duplicate allocations lead to memory leaks. Addresses #4709 [BFB] | 27 January 2022, 20:14:24 UTC |
| 3a1b5ed | noel | 27 January 2022, 16:20:21 UTC | Merge branch E3SM-Project/azamat/cori/fates-memleak-tolerance (PR #4739) Set memleak tolerance at 15% for FATES tests on Cori | 27 January 2022, 16:20:21 UTC |
| 9bd403b | James Foucar | 26 January 2022, 22:24:45 UTC | Merge branch 'bartgol/homme/cmake-changes' into master (PR #4749) Some minor CMake changes in Homme These changes only affect Homme standalone (since EAM does not use CMake). They allow a cmake host project to retain control of compiler flags, rather than having Homme decide. [BFB] * origin/bartgol/homme/cmake-changes: HOMME: removed HOMME_BUILD_LIBS cmake option, which was never used HOMME: handle -mkl flag on a per-target basis HOMME: set compiler flags only in standalone mode | 26 January 2022, 22:24:45 UTC |
| efc3089 | Eva Sinha | 26 January 2022, 16:40:29 UTC | Modifed parameters for miscanthus and switchgrass based on calibration and validation finalized for revised manuscript submitted in GMD https://gmd.copernicus.org/preprints/gmd-2021-244/ | 26 January 2022, 21:42:33 UTC |
| 0a64dc6 | Jon Wolfe | 26 January 2022, 20:25:20 UTC | Add support for the oRRS18to6v3 grid to run with JRA forcing | 26 January 2022, 20:25:20 UTC |
| e749872 | Gautam Bisht | 26 January 2022, 18:39:15 UTC | Merge branch 'aldivi/elm/update_fates_harvest' (PR #4024) ELM code (for harvest control and FATES harvest control) has been updated so that ELM input harvest can be passed to FATES to drive FATES harvest code. The ELM namelist variables do_harvest and fdyndat can now be used to drive FATES harvest. The FATES code has not been modified, as it has already been updated for this purpose. Modifications to the harvest read and get year functions were required to ensure that the correct harvest year data are read in. ELM harvest results are unchanged, and FATES harvest matches the input based on secondary forest fraction. Branch run control has been updated to enable branch runs. [non-BFB] only for FATES and BGC cases Conflicts: components/elm/src/main/elm_initializeMod.F90 components/elm/src/main/elmfates_interfaceMod.F90 components/elm/src/main/pftvarcon.F90 | 26 January 2022, 18:39:15 UTC |
| 22b70a6 | Eva Sinha | 23 September 2021, 17:29:41 UTC | Modifies perennial bioenergy crop harvest | 26 January 2022, 17:36:48 UTC |
| 020660a | Benjamin R. Hillman | 25 January 2022, 20:14:50 UTC | Merge branch 'whannah/mmf/add-amip-compset' (#4719) This adds a transient prescribed SST compset (i.e. AMIP) for the C++ MMF. We didn't bother with the fortran MMF because they can't run on GPUs, and therefore aren't good candidates for AMIP experiments. This also only implements the 1-moment microphysics case with prescribed aerosols as the new P3 implementation in the C++ MMF code is still a work in progress. Changes are BFB, but the ERS_Ln9.ne4pg2_ne4pg2.F-MMFXX.eam-mmf_use_VT test in e3sm_integration is replaced with ERS_Ln9.ne4pg2_ne4pg2.F20TR-MMFXX.eam-mmf_use_VT to add test coverage for the new compset, so a new baseline will need to be generated. [BFB] | 25 January 2022, 20:14:50 UTC |
| 816388d | Oksana Guba | 25 January 2022, 18:22:56 UTC | Merge branch 'origin/oksanaguba/hommexx/forcing-addon' (PR #4689) Enabling thetaxx as CAM_TARGET. Sorting out some of forcing logic for thetaxx in EAM (partially verified) and SCREAM (not verified and needs some sync with SCREAM repo). thetaxx does not work with threads. Adding an ERS test without threads. [bfb] but a new test. | 25 January 2022, 18:22:56 UTC |
| 01ba911 | Azamat Mametjanov | 24 January 2022, 22:48:27 UTC | Merge branch 'azamat/pgi/fix-openacc-tests' (PR #4748) Update PGI compiler runs of E3SM on GPUs: - add mpaso files with expanded OpenACC to Depends.pgigpu.cmake - fix missing CPP macro in threaded PGI runs - update and remove deprecated PGI/NVHPC flags - default ne30.pg2 runs to 2 nodes on Summit and Ascent Fixes E3SM-Project/E3SM#4363 [BFB] | 24 January 2022, 22:48:27 UTC |
| c8866c5 | Oksana Guba | 24 January 2022, 22:33:15 UTC | fix to finalize | 24 January 2022, 22:33:15 UTC |
| 91bfee5 | Matthew Hoffman | 17 November 2021, 22:22:54 UTC | Replace actualWhen with saveActualWhen This commit modified the previous commit by removing the actualWhen argument to the MPAS_stream_mgr_read() routine and replacing it with an optional argument saveActualWhen. The actualWhen argument was ambiguous when MPAS_stream_mgr_read was called in a way such that multiple streams were read - the actualWhen value returned came from the final stream read and could possibly be wrong for the other streams that were read. Given the previous commit added the mostRecentAccessTime attribute to the MPAS_stream_list_type type, the actualTime argument was somewhat redundant anyway. Now, instead of being able to have the actualTime returned as an argument, there is the option to have the actual time saved to the new attribute. It can then be retrieved from the stream cursor as needed. The saveActualWhen argument is optional because some streams do not contain xtime and trying to obtain ‘actual when’ in those cases will cause the reading of the stream to fail. Therefore, it is important to be able to not seek this information if possible. If the saveActualWhen argument is not provided to MPAS_stream_mgr_read(), its value defaults to false. This commit also changes the land ice core to stop using actualWhen and replace it with saveActualWhen. Neither the ocean nor seaice cores use actualWhen so no other changes are necessary. | 24 January 2022, 17:12:27 UTC |
| 91b290d | Matthew Hoffman | 17 November 2021, 18:11:42 UTC | Add mostRecentAccessTime attribute to streams This commit adds a mostRecentAccessTime attribute to all streams. For input, this corresponds to the time for which input was last read. For output, this corresponds to the time for which output was last written. Note that for output, there is an option to specify an output time that is not the current time; the value used here for mostRecentAccessTime uses the time actually sent to the file. This attribute is initialized to '9999-12-31_23:59:59' and will contain that value prior to the first access. | 24 January 2022, 17:12:16 UTC |
| edfa5d2 | Xylar Asay-Davis | 23 January 2022, 23:20:55 UTC | Initialize Coriolis param in its own subroutine In situations where we do not expand the Earth sphere radius, the Coriolist parameter still needs to get initialized to a realistic value. This merge creates a new subroutine for initializing the Coriolis parameter that is calle regardless of whether the Earth sphere is being expanded or not. | 23 January 2022, 23:59:00 UTC |
| 601e512 | Oksana Guba | 22 January 2022, 23:36:47 UTC | fix runs with debug flags, init was missing | 22 January 2022, 23:36:47 UTC |
| 2845fa7 | Oksana Guba | 22 January 2022, 23:36:10 UTC | fix preqx build | 22 January 2022, 23:36:10 UTC |
| b010d80 | Azamat Mametjanov | 22 January 2022, 17:04:36 UTC | Update pe-layout | 22 January 2022, 17:04:36 UTC |
| d13164a | Luca Bertagna | 21 January 2022, 22:56:17 UTC | HOMME: removed HOMME_BUILD_LIBS cmake option, which was never used | 21 January 2022, 22:56:17 UTC |
| ddc2f46 | Luca Bertagna | 21 January 2022, 22:23:06 UTC | HOMME: handle -mkl flag on a per-target basis | 21 January 2022, 22:23:06 UTC |
| 4cb90a2 | Luca Bertagna | 21 January 2022, 22:22:09 UTC | HOMME: set compiler flags only in standalone mode | 21 January 2022, 22:22:09 UTC |
| 3523b2b | noel | 21 January 2022, 21:21:35 UTC | Merge branch E3SM-Project/ndk/machinefiles/PM-module-fix (PR #4742) For Perlmutter, remove a line to load environment module that is no longer needed | 21 January 2022, 21:21:35 UTC |
| e5e8122 | Oksana Guba | 21 January 2022, 21:16:16 UTC | fix preqxx build | 21 January 2022, 21:16:16 UTC |
| 0329967 | Jon Wolfe | 21 January 2022, 20:05:23 UTC | Merge branch 'MALI-Dev/develop' (PR #4725) Update MALI version and Greenland mesh This merge improves support for a dynamic Greenland ice sheet in E3SM using MALI. This includes: * Updating the version of Albany used on Cori to allow velocity solver improvements * Update to the GIS 1km initial condition file * Removing unneeded land model warnings These changes have been tested in both IG and BG compsets and run for one year at 5 day timesteps and 5 day glc coupling intervals. Results were examined qualitatively and deemed to be reasonable and realistic. Other features not generally used in E3SM: * Change from using beta as basal friction field to muFriction * Improvements to calving and facemelting parameterizations that allow * for exploration of more sophisticated calving treatment (off in E3SM) * Separates a flood fill routine for general use (BFB under all configurations) * A new namelist option to restore thickness after advection needed for some idealized configurations (not used in E3SM) * Improvements and fixes to the subglacial hydrology model (not * currently used in E3SM) Bug fixes: * Effective viscosity calculation caused a divide by zero in some situations and is now fixed * Fix bug to ensure time-varying forcing is read on initial time * Missing error variables are fixed that led to improper error handling in some cases * Damage was not being advected - that is now fixed (damage evolution is not currently used in E3SM) [NML] [non-BFB] for configurations with MALI | 21 January 2022, 20:05:23 UTC |
| 411351f | Jon Wolfe | 21 January 2022, 20:00:25 UTC | Merge branch 'xylar/ocn/init/fix-60-vert-lev-phc' (PR #4729) Fix interface locations for 60-layer PHC grid The previous values were the layer centers, not the layer interfaces for all but the top layer interface. See MPAS-Dev/compass#298 for the bug report, since this issue affects standalone MPAS-Ocean (in init mode) and not E3SM directly. [BFB] | 21 January 2022, 20:00:25 UTC |
| e3e4fb9 | Azamat Mametjanov | 21 January 2022, 19:49:28 UTC | Fix missing CPP macro in threaded PGI runs | 21 January 2022, 19:49:28 UTC |
| 53cf0fd | Azamat Mametjanov | 21 January 2022, 19:45:29 UTC | Update and remove deprecated PGI/NVHPC flags on Summit | 21 January 2022, 19:45:29 UTC |
| aaef335 | Azamat Mametjanov | 21 January 2022, 19:34:50 UTC | Run ne30pg2 tests on 4 nodes on Summit+Ascent to avoid OOM errors | 21 January 2022, 19:34:50 UTC |
| c79d2b0 | James Foucar | 21 January 2022, 19:24:42 UTC | Merge branch 'singhbalwinder/machines/compy-typo-fix' into master (PR #4747) Fixes a typo in compy's intel cmake file This typo breaks model compilation in the debug mode on Compy Fixes #4746 [BFB] - Bit-For-Bit * origin/singhbalwinder/machines/compy-typo-fix: Fixes a typo in compy's intel cmake file | 21 January 2022, 19:24:42 UTC |
| 805d20f | Azamat Mametjanov | 21 January 2022, 18:58:56 UTC | Update PGI compiler runs of E3SM on GPUs | 21 January 2022, 18:58:56 UTC |
| 95a0076 | Balwinder Singh | 21 January 2022, 01:01:47 UTC | Fixes a typo in compy's intel cmake file This typo breaks model compilation in the debug mode on Compy Fixes #4746 [BFB] - Bit-For-Bit | 21 January 2022, 01:03:36 UTC |
| 203b02a | Matthew Hoffman | 20 January 2022, 19:38:43 UTC | Merge branch 'matthewhoffman/fix_cori_cmake_again' into develop This PR removes an inadvertent stray character. * matthewhoffman/fix_cori_cmake_again: Remove inadvertent stray character that was in previous PR | 20 January 2022, 19:38:43 UTC |
| 48ed1bb | Matthew Hoffman | 20 January 2022, 19:35:04 UTC | Remove inadvertent stray character that was in previous PR | 20 January 2022, 19:35:04 UTC |
| 7b97a89 | Peter Schwartz | 20 January 2022, 18:37:29 UTC | Merge remote-tracking branch 'glemieuxfork/fates-spmode-rgknox_cnp_base'(PR #4669) This pull request makes E3SM compatible with FATES API 17.0.0 which enables the use of the FATES satellite phenology mode (aka "sp mode"). While FATES SP mode is operating, the state variables of TLAI, TSAI, HTOP and WTCOL are manipulated (from the surface dataset files) in the native HLM PFT based index space. In this PR, we set up mapping from the FATES IFP space into the HLM P space in elmfates_interfaceMod.F90. Note that the TLAI, TSAI, and HTOP history outputs are updated with a logic check on use_fates_sp to substitute the HLM variable output with the relevant fates sp mode version of these variables, which are identified with the suffix _hist_patch. This PR also introduces FATES no competition mode (use_fates_nocomp), which along with fixed biogeography mode (use_fates_fixed_biogeog), are necessary dependencies for fates sp mode. Fixed biogeography has been enabled in the namelist definition (previously it was set as a stub). The SatellitePhenology procedure has been slightly refactored to represent that it accepts different filters depending on whether or not it is being called for FATES sp mode or ELM sp mode as the FATES sp mode implementation needs to have access to both the active and inactive soil filter. [nonBFB] FATES cases | 20 January 2022, 18:37:29 UTC |
| b396be8 | James Foucar | 20 January 2022, 17:43:33 UTC | Merge branch 'jgfouca/fix_mpas_cmake_missing_dep' into master (PR #4736) Fix missing dependency in one case of gen_f90_targets This bug was causing .F changes to not be reflected in rebuilds. Fixes #4341 [BFB] * jgfouca/fix_mpas_cmake_missing_dep: Fix missing dependency in one case of gen_f90_targets | 20 January 2022, 17:43:33 UTC |
| 4326367 | Balwinder Singh | 20 January 2022, 17:13:22 UTC | Merge remote-tracking branch 'mt5555/ninst-fixes' (PR #4654) Fix component namelist creation bugs when NINST>1 A few bug fixes when running with NINST>1, as is done in the E3SM nbfb test suite MPAS sea ice component should use NTASKS_PER_INST_ICE instead of NTASKS_ICE when determining which partition file to use Ensure ATM and LND put the initial conditions from all ensemble members into the inputdata file list Similar fixes should probably be applied to the MPAS ocean (NTASKS_PER_INST_OCN) and ensemble member inputdata parsing for both MPAS ocean and ice. Not done in this PR since we currently dont use this capability. fixes #4648, #4645 [BFB] * mt5555/ninst-fixes: fix component namelist creation bugs when NINST>1 | 20 January 2022, 17:13:22 UTC |
| e8cc524 | Jon Wolfe | 20 January 2022, 16:03:51 UTC | Merge branch 'jonbob/scripts/update-ocn-bgc' (PR #4686) Update BGC compsets to turn ocn bgc back on, add new ocn bgc init file Adds new ocn BGC files that are consistent with the current MARBL-based implementation. This allows the ocn BGC to be turned back on in BGC compsets. Fixes #4616 [non-BFB] only for BGC compsets | 20 January 2022, 16:03:51 UTC |
| 783a279 | Matthew Hoffman | 19 January 2022, 22:32:48 UTC | Merge branch 'matthewhoffman/fix_cori_cmake' into develop This was accidentally deleted in 66432b7. Note that I am restoring the version of this line currently on master that was updated Nov. 11, 2021, in ee9eabe. * matthewhoffman/fix_cori_cmake: Restore inadvertent deletion of FFLAGS from intel_cori-haswell.cmake | 19 January 2022, 22:32:48 UTC |
| 43d5559 | Matthew Hoffman | 19 January 2022, 22:27:04 UTC | Restore inadvertent deletion of FFLAGS from intel_cori-haswell.cmake This was accidentally deleted in 66432b7db334187c8b49c07312e5d58682806e53. Note that I am restoring the version of this line currently on master that was updated Nov. 11, 2021, in ee9eabe95ba8d69e4f7667d006bf1fa417325a24. | 19 January 2022, 22:27:04 UTC |
| 4d7f1e5 | Walter Hannah | 19 January 2022, 16:24:45 UTC | Update ROF_NCPL for testing AMIP compsets | 19 January 2022, 16:24:45 UTC |
| bfd9e0d | noel | 18 January 2022, 20:56:52 UTC | For PM, remove a line to load a module that is no longer needed. Only affects gpu builds. | 18 January 2022, 20:56:52 UTC |
| 528e96a | Gautam Bisht | 17 January 2022, 21:42:47 UTC | Merge branch 'singhbalwinder/atm/sfeng-cpl-BPRP-lnd-atm' (PR #4703) This PR adds two land-atmosphere compsets (1850 and 20TR). Smoke tests are also added for these compsets. [BFB] | 17 January 2022, 21:42:47 UTC |
| 694a0c8 | Oksana Guba | 17 January 2022, 20:58:06 UTC | Merge branch 'origin/master' into oksanaguba/hommexx/forcing-addon resolve conflict around nsplit Conflicts: components/homme/src/share/cxx/SimulationParams.hpp components/homme/src/theta-l_kokkos/cxx/cxx_f90_interface_theta.cpp components/homme/src/theta-l_kokkos/prim_driver_mod.F90 components/homme/src/theta-l_kokkos/theta_f2c_mod.F90 | 17 January 2022, 20:58:06 UTC |
| e0cd910 | Oksana Guba | 17 January 2022, 20:12:58 UTC | following AMB suggestions | 17 January 2022, 20:12:58 UTC |
| 0083189 | Oksana Guba | 17 January 2022, 02:42:09 UTC | remove build xx test | 17 January 2022, 02:42:09 UTC |
| 771bd91 | Oksana Guba | 17 January 2022, 02:31:12 UTC | adding ers test, no threads, passes | 17 January 2022, 02:31:12 UTC |
| 95c9859 | Oksana Guba | 17 January 2022, 01:57:05 UTC | removing 0/0 for padding | 17 January 2022, 01:57:05 UTC |
| 25f515f | Oksana Guba | 17 January 2022, 01:47:20 UTC | cleanup | 17 January 2022, 01:47:20 UTC |
| 7976579 | Oksana Guba | 17 January 2022, 01:31:01 UTC | put a stop for xx dycore + omp | 17 January 2022, 01:31:01 UTC |
| 5de0be0 | Mark Taylor | 14 January 2022, 17:34:18 UTC | Merge branch 'mt5555/tom_sponge_start' (PR #4723) Add a simplified sponge layer with namelist variable (tom_sponge_start) to set the starting layer This is needed to run at high resolution with the E3SM 72L configuration (72L sponge layer is too thin, it starts at the 1mb layer). 72L configurations will crash at NE256 and higher resolution without making the sponge layer thicker. It will also be useful for the SCREAM 128L configuration, where the sponge layer is most likely too thick (it starts at the 20mb layer) default: tom_sponge_start=0 (disabled). documentation: https://acme-climate.atlassian.net/wiki/spaces/DOC/pages/2967798203/EAM+Top+of+Model+Sponge+Layer [BFB] | 15 January 2022, 17:03:14 UTC |
| b77e4b7 | Mark Taylor | 14 January 2022, 17:35:58 UTC | Merge branch 'brhillman/eam/fix-adjust-ps-scream' (PR #4717) Make setting of adjust_ps consistent with SCREAM repo. We want to allow adjust_ps=.false. for SCREAM, but retain adjust_ps=.true. for E3SM to remain BFB for v2. This change was originally made downstream in SCREAM, but causes conflicts between the two repos, and prevents EAM compsets from being BFB between the E3SM and SCREAM repos. Moving this change upstream, protected in an ifdef, allows the two repos to be consistent, and keeps EAM compsets BFB between the SCREAM and E3SM repos. This also sets the precedent for allowing changes upstream in E3SM that should apply only to SCREAM configurations, protected in ifdef SCREAM. To enable, we simply need to add the -cppdefs '-DSCREAM' flag to CAM_CONFIG_OPTS for SCREAM compsets. [BFB] | 15 January 2022, 17:01:56 UTC |
| 80d9df9 | Azamat Mametjanov | 15 January 2022, 00:10:39 UTC | Set TEST_MEMLEAK_TOLERANCE=15% for FATES tests on Cori | 15 January 2022, 00:10:39 UTC |
| 0a3d404 | Azamat Mametjanov | 15 January 2022, 00:07:18 UTC | Move MEMLEAK+TPUT tolerance to env_run.xml for testmods | 15 January 2022, 00:07:18 UTC |
| 58826d8 | Mark Taylor | 14 January 2022, 17:32:27 UTC | Merge branch 'mt5555/imex-unit-tests' into master (PR #4731) turn on (during startup) IMEX test of analytic jacobian ( verify that it it agrees with numerical approximation) turn on (during startup) test of the A. Bradley tridiagonal solver [BFB] | 14 January 2022, 22:07:48 UTC |
| 8441102 | Jon Wolfe | 14 January 2022, 15:53:45 UTC | Merge branch 'jonbob/ocn/fix-bulk-forcing' (PR #4733) Make surfaceTemperatureFluxWithoutRunoff private in omp pragma This PR adds surfaceTemperatureFluxWithoutRunoff to an omp pragma to be private for a loop in the mpas-ocean routine ocn_surface_bulk_forcing_active_tracers. It was causing some tests to be irreproducible when run with threading on. Fixes #4732 [non-BFB] for some threaded tests | 14 January 2022, 15:53:45 UTC |
| ce6b973 | Gregory Lemieux | 13 January 2022, 22:02:59 UTC | Updating assignment to fates pft_areafrac If not using a surface dataset with topounits, only access the first index for the topounit dimension as max_topounit is defined in these cases to be one. Otherwise we uses the column type indexing. | 13 January 2022, 22:02:59 UTC |
| f6bb0c5 | Oksana Guba | 13 January 2022, 21:17:53 UTC | Merge branch 'origin/oksanaguba/homme/bubble-rjmods' into next (PR #4482) This fixes an oversaturated state in the moist bubble initialization. Also, qsat in initialization now depends on RJ or Kessler physics. [nonbfb] for 1 homme test, moist bubble. | 13 January 2022, 21:17:53 UTC |
| 75ec9db | James Foucar | 13 January 2022, 19:53:26 UTC | Fix missing dependency in one case of gen_f90_targets This bug was causing .F changes to not be reflected in rebuilds. [BFB] | 13 January 2022, 19:53:26 UTC |
| 38e3e36 | Benjamin R. Hillman | 13 January 2022, 19:12:45 UTC | Fix resolution in F20TR-MMFXX test | 13 January 2022, 19:12:45 UTC |
