d36c7bc | Jon Wolfe | 25 April 2022, 19:35:42 UTC | Merge branch 'sbrus89/ocn/add-self-attraction-loading' (PR #4472) Add in-line self attraction and loading for global tidal simulation Self attraction and loading (SAL) forcing involves a spherical harmonic transform (SHT) of the SSH (Ray, 1998). The spherical harmonic coefficients are scaled by the load Love numbers, then an inverse transformation is performed to compute the SAL field. The SAL forcing is incorporated in the pressure gradient term along with the tidal potential. Two different approaches are implemented for performing the SHT: 1. SSH is gathered onto a single MPI rank and interpolated to a Gaussian mesh. A very fast library, SHTns, is used to compute the forward/inverse SHT. The SAL field is then interpolated back to the unstructured mesh and scattered back to the other MPI ranks. This work was done by @knbarton. 2. The SHT integration is performed on each subdomain and the SH coefficients are globally summed with an all-reduce. Approach 1 ) performs better at low core-counts, but 2) scales better for larger runs. This PR includes an init mode option for verification testing of the forward/inverse SHT. This work was done as a part of the tides task of the ICoM project. [NML] [BFB] for all standard E3SM tests | 25 April 2022, 19:35:42 UTC |
cebe3f9 | James Foucar | 25 April 2022, 19:25:02 UTC | Merge branch 'jgfouca/cime_ninja_update' into master (PR #4911) Update CIME by one commit Needed to fix get_case_env [BFB] | 25 April 2022, 19:25:02 UTC |
f92dd7e | James Foucar | 25 April 2022, 19:22:38 UTC | Update CIME by one commit Needed to fix get_case_env [BFB] | 25 April 2022, 19:22:38 UTC |
3bead91 | Azamat Mametjanov | 23 April 2022, 02:38:09 UTC | Merge branch 'azamat/chrys/throttle-bld-par' (PR #4909) Reduce parallel build jobs on Chrysalis down to 4 to reduce intel license errors. [BFB] | 23 April 2022, 02:38:09 UTC |
f7950c8 | Azamat Mametjanov | 23 April 2022, 02:21:47 UTC | Reduce parallel build jobs on Chrysalis down to 4 | 23 April 2022, 02:21:47 UTC |
284527b | Jon Wolfe | 21 April 2022, 17:56:32 UTC | Remove extraneous character | 21 April 2022, 17:56:32 UTC |
abfdea1 | Jon Wolfe | 21 April 2022, 17:30:08 UTC | Add updates missed in mpas-ocean build-namelist file | 21 April 2022, 17:30:08 UTC |
a376631 | Oksana Guba | 20 April 2022, 22:16:44 UTC | Merge 'bartgol/homme/prevent-full-rebuild-due-to-sha-change' into next (PR #4889) Currently, in homme standalone mode, the GIT sha of the repo is stored in the config.h files. This means that if one adds a commit, and/or rewrites history, the sha changes. Since these files are included (directly or indirectly) by _a lot_ of other files in Homme, this cause basically a full rebuild of Homme. This PR allows to do changes that affect the git sha (such as amending a commit) without triggering a full rebuild of all execs/libs. I verified that the only files to be rebuilt are all the copies of namelist_mod, which is the only place that uses the GIT sha (printing it to screen for debug purposes). Side mod: when the var CPRNC_DIR is not set, Homme set it to the internal path to CIME's cprnc. However, recent changes in CIME changed the location of cprnc, so this path had to be adjusted. This only affects builds where the CPRNC_DIR var is _not_ set. If a mach file already stores CPRNC_DIR, nothing changes with this PR. [bfb] | 20 April 2022, 22:16:44 UTC |
1972b36 | Jon Wolfe | 20 April 2022, 17:19:34 UTC | Merge branch 'njeffery/seaice/update_carbon_analysis_member' (PR #4826) Update carbon conservation analysis member Brings in an updated seaice carbon conservation am that: * tracks individual ocean-ice carbon fluxes * computes accumulated error, relative errors, error bounds over the run * adds flag option to end the run if carbon errors exceed bounds * includes three additional diagnostics * adds molecular weight of carbon as a shared constant for all carbon conservation calculations * includes a non-BFB correction in merge-fluxes-bgc which uses the initial ice area to merge BGC fluxes rather than the ice area after step therm 1. Passes all tests in the sea ice testing suite. [NML] [non-BFB] only for runs with active sea ice BGC | 20 April 2022, 17:19:34 UTC |
083ccd2 | jayeshkrishna | 19 April 2022, 21:02:06 UTC | Merge branch 'dqwu/anlgce/update-modules' (PR #4873) Most GCE nodes have been upgraded from Ubuntu 18.04 to 20.04 and we need to update some modules and software libs accordingly to run E3SM on these nodes. A new machine name anlgce-ub18 is created to support two legacy GCE nodes (compute-01.cels.anl.gov and compute-02.cels.anl.gov) that still use Ubuntu 18.04. Open MPI support is also added for ANL GCE nodes in this PR. [BFB] | 19 April 2022, 21:02:06 UTC |
48e9195 | Peter Schwartz | 19 April 2022, 17:33:45 UTC | Merge branch 'jayeshkrishna/mosart_mean_mnth_flow_shrt_ckt_fix' (PR #4861) Since the Fortran standard does not support short circuit evaluation, avoiding indexing the array when the index can be <= 0. Fixes #4727 [BFB] | 19 April 2022, 17:33:45 UTC |
bfcfece | Hyun-Gyu Kang | 19 April 2022, 15:18:14 UTC | Change a reproducible sum algorithm in SI solver Change a reproducible sum algorithm in the ocean split-implicit solver. The changed algorithm uses a global reproducible summation module in components/mpas-framework/src/framework/mpas_global_sum_mod.F | 19 April 2022, 15:18:14 UTC |
e3098ee | Jason Sarich | 19 April 2022, 15:14:40 UTC | Merge branch 'azamat/chrys/update-openmpi-4.1.3' (PR #4890) Update Chrysalis to OpenMPI-4.1.3 Addresses #4819 [BFB] | 19 April 2022, 15:14:40 UTC |
87364c9 | Benjamin R. Hillman | 15 April 2022, 17:34:52 UTC | Merge branch 'whannah/mmf/fix-m2005+ECPP' (#4846) The 2-moment configuration of MMF currently is not tested because we don't plan to use it, but it will be nice to have when enabling the MMF to use P3+ECPP that is planned. This also includes a cosmetic rewrite of the crmclouds_camaerosols module. [BFB] - because the broken compset is not tested, which is ok | 15 April 2022, 17:34:52 UTC |
7d6d3d2 | Azamat Mametjanov | 15 April 2022, 00:16:22 UTC | Update Chrysalis to OpenMPI-4.1.3 | 15 April 2022, 00:16:22 UTC |
4435bfa | Peter Schwartz | 14 April 2022, 20:34:19 UTC | Merge remote-tracking branch 'rgknoxfork/rgknox/lnd/fates-conditional-timesteps'(PR #4884) This is a simple fix to prevent FATES code that handles some time keeping routines, from being called when FATES is not active. These routines should have no effect on a non-fates run. Fixes: #1537 Collaborators: AaronDonahue | 14 April 2022, 20:34:19 UTC |
c9cb519 | Azamat Mametjanov | 14 April 2022, 19:38:51 UTC | Merge branch 'azamat/pgi/update-flags' (PR #4863) Update PGI runs and pelayouts - add -O0 -g to all flags in debug runs - run kokkos --with-serial on CPU - add -Mnovect to eam and cice on Summit - reduce optimization on component_mod and cime_comp_mod - update to nvhpc/21.11 on Summit Ascent - add more OpenACC-enabled files - cleanup pgi cmake - run BGC cases with 4x42x1 pelayout - remove elm-solar_rad NTASKS=48 testmod in favor of default pelayout [NML] | 14 April 2022, 19:38:51 UTC |
35604de | Azamat Mametjanov | 14 April 2022, 19:14:41 UTC | Merge branch 'azamat/fugaku/initial-configure' (PR #4617) Add Fugaku machine configuration with gnu/11.2.0 and fujitsu/4.7.0 compilers. [BFB] | 14 April 2022, 19:14:41 UTC |
8aef31a | Jon Wolfe | 14 April 2022, 19:10:21 UTC | Merge branch 'mark-petersen/ocean/eddy-length-limiter-visbeck' (PR #4868) Limit Visbeck eddy length to dcEdge When using config_GM_closure='Visbeck' (Visbeck et al 1997), limit the eddy length scale to be no less than dcEdge, the distance between cells. This ensures that gmBolusKappa does not have very small values due to the eddyLength factor. When using config_GM_closure='Visbeck' it is not BFB. However, all E3SM runs use config_GM_closure='constant' or 'EdenGreatbatch', but never Visbeck, so this PR is BFB for all current tests. [BFB] | 14 April 2022, 19:10:21 UTC |
ae09450 | Azamat Mametjanov | 14 April 2022, 16:40:48 UTC | Merge branch 'azamat/anvil/bgc-wcycl-disambig' (PR #4886) Add default 30-node BGC pelayout on Anvil. Also disambiguate pelayouts for - WCYCL and BGC compsets - BGC compset with MPASO and DOCN components [BFB] | 14 April 2022, 16:40:48 UTC |
5ce3f84 | James Foucar | 14 April 2022, 16:30:05 UTC | Merge branch 'jgfouca/emergency_cime_fix' into master (PR #4888) Emergency CIME fix Do not probe machine and then raise an error, just use what user asked for. [BFB] * jgfouca/emergency_cime_fix: Emergency CIME fix | 14 April 2022, 16:30:05 UTC |
ba05d48 | Luca Bertagna | 14 April 2022, 01:06:08 UTC | Homme: move SHA info in separate header Avoids full project rebuild (due to config.h change) when checking out different versions of the code. A change in sha now only triggers the rebuild of namelist_mod, and only for standalone Homme builds. | 14 April 2022, 01:50:49 UTC |
9e75da5 | Luca Bertagna | 14 April 2022, 01:20:11 UTC | Homme: fix path to cprnc, following recent CIME update | 14 April 2022, 01:50:46 UTC |
802b189 | James Foucar | 13 April 2022, 22:21:33 UTC | Emergency CIME fix Do not probe machine and then raise an error, just use what user asked for. [BFB] | 13 April 2022, 22:21:33 UTC |
85d338d | Azamat Mametjanov | 13 April 2022, 17:08:15 UTC | Merge branch 'azamat/tests/rm-mmf-omp-test' (PR #4879) Update OpenMP-offload tests: - rm SMS_Ln9.ne4pg2_ne4pg2.F-MMFOMP.eam-single_thread from e3sm_gpuomp - adjust Miller NODENAME_REGEX to avoid login node name conflicts [BFB] | 13 April 2022, 17:08:15 UTC |
65e1769 | James Foucar | 13 April 2022, 16:24:49 UTC | Merge branch 'jgfouca/cime_update_2022_04_07' into master (PR #4877) Update CIME submodule ... to 524842660578d7b8f72df78a77bd302f9d128b44 Changes: 1) Make sure CIME machine matches probed machine 2) Update neon data server for input files 3) New SystemTestsCompareN test type 4) Extend configure to get cmake macros files from the .cime directory 5) BIG: restructure CIME to match what would be expected in a standard python pacakge! 6) wait_for_tests: Try multiple ctest drop methods 7) Allow special XML syntax for init_path and cmd_path 8) Add Total_Build time to end of build_times.txt Fixes: 1) Fix the path for PTS_DOMAIN_FILE 2) Fix config_machines xsd to allow more compilers 3) Fix xmlchange tests 4) Remove Machines warning 5) Using --retry in create_newcase was causing generated baseline to be removed 6) Fix run_tests.py when no machine is given 7) Fixes for machine check 8) Fixes issue with model detection on E3SM machines. Fixes #4875 [BFB] * jgfouca/cime_update_2022_04_07: E3SM-side fixes needed Update CIME submodule | 13 April 2022, 16:24:49 UTC |
77c2ded | Jon Wolfe | 13 April 2022, 16:09:30 UTC | Merge branch 'philipwjones/mpas-ocean/acc-thickfix' (PR #4866) Fixes OpenACC bugs introduced by recent layerThickEdge changes During a recent addition of layerThickEdgeFlux/Mean variables, some GPU errors were introduced: * layerThickEdgeXXX vars were missing from some OpenACC update directives * the use of array syntax when layerThickEdgeFlux set to layerThickEdgeMean was causing the operation to happen on host (with bad data) rather than device when OpenACC turned on * replaced complicated string conditionals with case construct to improve performance and clarity * the conditionals above were moved to diagnostics init * cleaned up the thickness edge routine a little * fixed related OpenACC present data issues in surface pressure calculation by moving option conditionals outside loop and splitting loop accordingly Tested on Summit w/ PGI in QU240 and bfb when OpenACC turned on Fixes #4865 [BFB] | 13 April 2022, 16:09:30 UTC |
4af11ef | Jon Wolfe | 13 April 2022, 15:23:50 UTC | Merge branch 'jonbob/coupler/fix-trigrid-lnd2atm' (PR #4862) Fix the lnd to atm direct fluxes for trigrid configurations The process of creating a carbon budget pointed out that we have to been handling the "direct" lnd-to-atm fluxes incorrectly for trigrid configurations. The new budget will come in separately, but this brings in the coupler changes as well as a consistent update to the heat and energy budgets. [non-BFB] for trigrid simulations | 13 April 2022, 15:23:50 UTC |
a40e35b | Azamat Mametjanov | 13 April 2022, 03:18:51 UTC | Add small BGC pelayout on Anvil Also disambiguate pelayouts for - WCYCL and BGC - BGC with MPASO and DOCN | 13 April 2022, 03:18:51 UTC |
98ff66a | Jon Wolfe | 12 April 2022, 22:17:47 UTC | Update documentation | 12 April 2022, 22:17:47 UTC |
8d267a6 | Wuyin Lin | 12 April 2022, 21:23:01 UTC | Merge 'wlin/lnd/revert_ne30pg2_defaults' (PR #4881) Revert changes to elm namelist_defaults for ne30pg2 and simyr1850 Revert the changes to fsurdat and flndtopo for ne30pg2 and 1850. The changes were introduced by #4781. [BFB] but NML diff when involving 1850 land on ne30pg2 | 12 April 2022, 21:23:01 UTC |
dca950e | Jon Wolfe | 12 April 2022, 21:06:36 UTC | Update budget table format | 12 April 2022, 21:06:36 UTC |
3851c52 | Ryan Knox | 12 April 2022, 18:55:29 UTC | Adding conditional use_fates to prevent entering FATES routine for setting some averaging info. This was detected by AaronDonahue in issue 1537 | 12 April 2022, 18:55:29 UTC |
a49744f | James Foucar | 12 April 2022, 17:03:42 UTC | Fix circle CI | 12 April 2022, 17:03:42 UTC |
d20c23f | James Foucar | 12 April 2022, 16:56:43 UTC | Fix homme timing dirs | 12 April 2022, 16:56:43 UTC |
570bd47 | Wuyin Lin | 12 April 2022, 16:39:26 UTC | Revert changes to elm namelist_defaults introduced by #4781 | 12 April 2022, 16:39:26 UTC |
48ff595 | Wuyin Lin | 12 April 2022, 15:00:31 UTC | Merge branch 'benedict/atm/ssp370_ssp585' (PR #4844) Add and update SSP370 and SSP585 Added SSP370 compset "WCYCLSSP370" and updated existing compset "WCYCLSSP585". For SSP370, the new "WCYCLSSP370" compset was defined, new EAM and ELM use case files were created, new prescribed forcing files were created, and minor changes elsewhere were added by emulating what was done for SSP585. For the updated SSP585, the "WCYCLSSP585" compset was defined, new land use and land cover change files were created (these files are grid-specific, and the default atm_lnd grid changed from v1 to v2), and the ELM use case file was updated to point to these new files. The atmosphere forcing files set in the SSP585 use case were not changed because they are not grid specific and they already existed from v1. All SSP370 EAM and ELM input files were staged in NERSC-Cori's inputdata directory only. [BFB] for tests not using SSP370 or SSP585 | 12 April 2022, 15:00:31 UTC |
175426d | Jon Wolfe | 11 April 2022, 21:08:42 UTC | Convert CO2 fluxes to carbon in the budget | 11 April 2022, 21:08:42 UTC |
95fbb5c | Benjamin Hillman | 11 April 2022, 20:27:38 UTC | Merge remote-tracking branch 'xyuan/xyuan/samxx_p3_shoc' (#4831) using YAKL_SCOPE to replace the auto to create local variables that required by SYCL on JLSE [BFB] | 11 April 2022, 20:27:38 UTC |
f1886bc | Jon Wolfe | 06 April 2022, 21:50:23 UTC | Add BGC budget terms to coupler restart files | 11 April 2022, 19:03:03 UTC |
733b576 | Jon Wolfe | 06 April 2022, 21:49:35 UTC | Move seq_diagBGC_preprint_mct call, so it actually works | 11 April 2022, 19:02:37 UTC |
b77c1c7 | Jon Wolfe | 24 March 2022, 22:18:55 UTC | Add do_bgc_budgets flag | 11 April 2022, 19:01:54 UTC |
a48e798 | Jon Wolfe | 11 April 2022, 18:59:51 UTC | Make lnd fraction dependent on samegrid_al | 11 April 2022, 18:59:51 UTC |
bbd4b6f | Jon Wolfe | 28 February 2022, 19:53:29 UTC | Initial implementation of carbon budget in the coupler | 11 April 2022, 18:50:08 UTC |
31570d5 | Balwinder Singh | 11 April 2022, 17:56:27 UTC | Merge branch 'singhbalwinder/mct/add-prnt-out-bfbflag'... (PR #4802) Adds print statement to add a print out of BFBFLAG * singhbalwinder/mct/add-prnt-out-bfbflag: Adds BFBFLAG prinouts to cpl history file Moves bfbfflag print logic to init2 Adds print statement to add a print out of BFBFLAG | 11 April 2022, 17:56:27 UTC |
d10b334 | James Foucar | 11 April 2022, 16:01:12 UTC | CIME path fixes | 11 April 2022, 16:01:12 UTC |
8a76724 | Azamat Mametjanov | 09 April 2022, 00:53:31 UTC | Rm SMS_Ln9.ne4pg2_ne4pg2.F-MMFOMP.eam-single_thread from e3sm_gpuomp Also adjust Miller NODENAME_REGEX to avoid login node name conflicts | 09 April 2022, 00:53:31 UTC |
deb2244 | Azamat Mametjanov | 09 April 2022, 00:31:43 UTC | Run BGC cases at 4x42x1 | 09 April 2022, 00:31:43 UTC |
98703cb | Azamat Mametjanov | 09 April 2022, 00:29:01 UTC | Remove elm-solar_rad NTASKS=48 testmod in favor of default pelayout | 09 April 2022, 00:29:01 UTC |
22db376 | Min Xu | 08 April 2022, 17:22:00 UTC | Merge branch 'minxu74/machinefiles/miller' (PR #4781) The machine settings are for the US Air Force two HPE Cray EX supercomputers (miller and fawbush) at ORNL. Each system consists of 800 computer nodes. Each node has two 2nd Gen 7742 Epyc 64-core AMD Rome Processors and 256 GB memory. Nodes are connected via the 100GB "Slingshot". It also includes 20 GPU nodes on miller supercomputer and each node contains four Nvidia A100 GPUs. Three compilers are supported on Miller, GNU, Nvidia/nvhpc and AMD/aocc. | 08 April 2022, 17:22:00 UTC |
9f59ec6 | Azamat Mametjanov | 08 April 2022, 00:29:37 UTC | Replace cmake set with append | 08 April 2022, 00:29:37 UTC |
4f6257c | Azamat Mametjanov | 07 April 2022, 23:38:47 UTC | Upgrade to nvhpc/21.11 on Summit and Ascent | 07 April 2022, 23:38:47 UTC |
7dfd4db | Azamat Mametjanov | 07 April 2022, 23:27:09 UTC | Clean up PGI cmake | 07 April 2022, 23:27:09 UTC |
08b134c | James Foucar | 07 April 2022, 21:55:25 UTC | E3SM-side fixes needed | 07 April 2022, 21:55:25 UTC |
808a2bd | James Foucar | 07 April 2022, 20:07:02 UTC | Update CIME submodule ... to 524842660578d7b8f72df78a77bd302f9d128b44 Changes: 1) Make sure CIME machine matches probed machine 2) Update neon data server for input files 3) New SystemTestsCompareN test type 4) Extend configure to get cmake macros files from the .cime directory 5) BIG: restructure CIME to match what would be expected in a standard python pacakge! 6) wait_for_tests: Try multiple ctest drop methods 7) Allow special XML syntax for init_path and cmd_path 8) Add Total_Build time to end of build_times.txt Fixes: 1) Fix the path for PTS_DOMAIN_FILE 2) Fix config_machines xsd to allow more compilers 3) Fix xmlchange tests 4) Remove Machines warning 5) Using --retry in create_newcase was causing generated baseline to be removed 6) Fix run_tests.py when no machine is given 7) Fixes for machine check 8) Fixes issue with model detection on E3SM machines. [BFB] | 07 April 2022, 20:07:02 UTC |
8d33109 | Azamat Mametjanov | 07 April 2022, 19:43:35 UTC | Merge branch 'azamat/eam-chem-pp/dont-bld-with-xlf95' (PR #4740) Fix EAM Chem PP built with IBM compiler Also, do not copy campp preprocessor exe to components/eam/chem_proc. Fixes E3SM-Project/E3SM#4499 [BFB] | 07 April 2022, 19:43:35 UTC |
90ff912 | Jon Wolfe | 07 April 2022, 17:03:57 UTC | Update from scripts that autmoate bld files from Registry changes | 07 April 2022, 17:03:57 UTC |
543ccf3 | Azamat Mametjanov | 06 April 2022, 22:20:22 UTC | Simplify cmake conditionals | 06 April 2022, 22:20:22 UTC |
ea12e8d | Azamat Mametjanov | 24 February 2022, 00:48:31 UTC | Update kokkos module with Fujitsu compiler mods | 06 April 2022, 21:56:39 UTC |
cc65163 | Azamat Mametjanov | 24 February 2022, 00:41:37 UTC | Update ELM-SBETR module hash with Fujitsu compiler mods | 06 April 2022, 21:56:39 UTC |
3241ae6 | Azamat Mametjanov | 24 February 2022, 00:38:58 UTC | Update spack modules and add fujitsu compiler mods | 06 April 2022, 21:56:03 UTC |
c0bc7d9 | Azamat Mametjanov | 05 November 2021, 00:17:33 UTC | Make Fujitsu-MPI the default MPI library | 06 April 2022, 21:53:10 UTC |
121d677 | Azamat Mametjanov | 30 October 2021, 02:35:20 UTC | Add perl, cmake and lapack modules | 06 April 2022, 21:53:09 UTC |
4f7a61b | Azamat Mametjanov | 22 October 2021, 02:04:51 UTC | Add initial machine config for Fugaku | 06 April 2022, 21:52:49 UTC |
5661f3d | dqwu | 06 April 2022, 16:52:43 UTC | Adding support for the Open MPI library on anlgce and anlgce-ub18 The following software packages are built statically with gcc 11.1.0 and Open MPI 4.1.3, * HDF5 1.12.1 * NetCDF C 4.8.0 * NetCDF Fortran 4.5.3 * PnetCDF 1.12.2 | 06 April 2022, 18:54:18 UTC |
682adde | Jon Wolfe | 06 April 2022, 16:46:57 UTC | Merge branch 'jonbob/scripts/fix-cori-trigrid' (PR #4872) Specify EAM in compset to avoid PE layout matching issue on cori Fixes a bug where a new default trigrid layout caused an error on cori-knl due to "more than one PE layout matches given PE specs" by specifying that EAM is part of the corresponding compset. Fixes #4870 [BFB] | 06 April 2022, 16:46:57 UTC |
565da94 | Jon Wolfe | 06 April 2022, 16:30:21 UTC | Merge branch 'mark-petersen/ocn/gm-min-phase-speed' (PR #4835) Add choice of computed or constant ocean GM gravity wave speed Currently the internal gravity wave speed (first baroclinic mode) in the ocean GM parameterization is a constant, previously named config_GM_constant_gravWaveSpeed but changed in this PR to config_GM_constant_bclModeSpeed. This remains the default, but this PR adds a new flag, config_GM_minBclModeSpeed_method: If 'constant' (the default) then use config_GM_constant_bclModeSpeed. If 'computed' then compute the gravity wave speed at every edge at every time step using the Brunt-Vaisala frequency. This functionality was tested by the AMOC focus team in 2021 on the branch amoc-experimental-branch. [NML] [BFB] | 06 April 2022, 16:30:21 UTC |
ad7fb32 | dqwu | 06 April 2022, 16:25:36 UTC | Adding back support for ANL GCE nodes that still use Ubuntu 18.04 Create a new machine name anlgce-ub18 for two legacy ANL GCE nodes that still use Ubuntu 18.04. This temporary support will be removed after these legacy nodes are upgraded to Ubuntu 20.04 or higher in the future. | 06 April 2022, 16:25:36 UTC |
4a3c8db | dqwu | 06 April 2022, 16:04:52 UTC | Updating some modules and libs for ANL GCE nodes Most (11 out of 13) ANL GCE compute nodes have been upgraded from Ubuntu 18.04 to Ubuntu 20.04 and we need update some modules and software libs accordingly to run E3SM on these nodes. | 06 April 2022, 16:04:52 UTC |
08cb673 | Azamat Mametjanov | 05 April 2022, 19:30:13 UTC | Limit EAM tri-grid pelayout to Anvil and Bebop Also specify full grid name for - SMS_Ln5.ne30pg2_r05_EC30to60E2r2.BGCEXP_LNDATM_CNPRDCTC_1850 - SMS_Ln5.ne30pg2_r05_EC30to60E2r2.BGCEXP_LNDATM_CNPRDCTC_20TR | 05 April 2022, 19:30:13 UTC |
cd998fe | Jon Wolfe | 05 April 2022, 19:05:44 UTC | Specify EAM for compset to avoid PE layout matching issue on cori | 05 April 2022, 19:05:44 UTC |
c99879a | Mark Petersen | 05 April 2022, 18:34:46 UTC | move lt1,lt2 initialization within loop for threading | 05 April 2022, 18:34:46 UTC |
0869d61 | Mark Petersen | 05 April 2022, 18:30:39 UTC | change eddyLength & eddyTime to scalars | 05 April 2022, 18:30:39 UTC |
ac5cf5a | Phil Jones | 05 April 2022, 16:01:59 UTC | temporary removal of ice pressure from present clause this is causing runs to fail with data not present, probably due to re-retrieval of pointer so is losing the device pointer. Will have a fix later but this is needed to run now. | 05 April 2022, 16:01:59 UTC |
03a5551 | Mark Petersen | 05 April 2022, 15:27:29 UTC | Update acc pragmas from Phil's comments | 05 April 2022, 15:27:29 UTC |
46dbaa4 | Azamat Mametjanov | 05 April 2022, 02:15:58 UTC | Reduce optimization for cime_comp_mod.F90 | 05 April 2022, 02:15:58 UTC |
64ad3f3 | Mark Petersen | 04 April 2022, 21:43:35 UTC | limit Visbeck eddy length to dcEdge | 04 April 2022, 21:43:35 UTC |
6d58bd6 | Phil Jones | 04 April 2022, 21:18:08 UTC | temporary removal of iceDraft from present clause although correct, the use of iceDraft in an OpenACC present clause was causing problems, possibly due to it being an optional argument, so removed for now and a fix will be introduced later | 04 April 2022, 21:18:08 UTC |
e5406e7 | Jason Sarich | 04 April 2022, 17:26:45 UTC | Merge branch 'azamat/pelayouts/anvil-trigrid-eam-4to8-node-update' (PR #4852) Update pelayouts for EAM tri-grid tests running into OOM errors on 4 Anvil nodes fixes tests: SMS_Ln5.ne30pg2_r05_EC30to60E2r2.BGCEXP_LNDATM_CNPRDCTC_1850.anvil_gnu SMS_Ln5.ne30pg2_r05_EC30to60E2r2.BGCEXP_LNDATM_CNPRDCTC_20TR.anvil_gnu [BFB] | 04 April 2022, 17:26:45 UTC |
6e0eed1 | jayeshkrishna | 04 April 2022, 16:51:10 UTC | Merge branch 'jayeshkrishna/scorpio_v1.3.1' (PR #4858) Updating to SCORPIO 1.3.2 SCORPIO 1.3.1 includes the following changes, * Improvements to the I/O performance statistics summary * Upgraded ADIOS file format from BP3 to BP4 * The internal GPTL library is now uptodate and synced with the version in E3SM * Removed upper limits on I/O descriptor ids * Added padding to all NetCDF output files to improve E3SM post processing performance * Misc updates to CMake configure * More timers in the Fortran interface * Support for gnu 11 * Misc bug fixes SCORPIO 1.3.2 includes the following changes, * Workaround for stale *loc* files on Chrysalis after E3SM runs * Workaround for E3SM crashes with the NVHPC (PGI) compiler on Summit/Ascent * Workaround for configure issue with ADIOS 2.8.0 (with SST) [BFB] | 04 April 2022, 16:51:10 UTC |
0195995 | Jon Wolfe | 04 April 2022, 16:12:24 UTC | Update bld files with output from scripts to match changes to Registry | 04 April 2022, 16:12:24 UTC |
e1e6473 | Phil Jones | 04 April 2022, 15:45:19 UTC | fix to manage openacc present data in ocn surface pressure calculation to manage data availability for various surface pressure forcing options, the loop needed to be split and conditionals for the various options moved outside the loop so the present clauses could accurately reflect the data used | 04 April 2022, 15:45:19 UTC |
58298dd | jayeshkrishna | 01 April 2022, 23:25:14 UTC | Upgrading to SCORPIO 1.3.2 Upgrading SCORPIO from 1.3.1 to 1.3.2 This patch includes the following changes (after 1.3.1), * Workaround for stale *loc* files on Chrysalis after E3SM runs * Workaround for E3SM crashes with the NVHPC (PGI) compiler on Summit/Ascent * Workaround for configure issue with ADIOS 2.8.0 (with SST) | 01 April 2022, 23:25:14 UTC |
f47e63c | Phil Jones | 01 April 2022, 22:59:08 UTC | fixes OpenACC bugs introduced by recent layerThickEdge changes - layerThickEdgeXXX vars were missing from some OpenACC update directives - the use of array syntax when layerThickEdgeFlux set to layerThickEdgeMean was causing the operation to happen on host (with bad data) rather than device when OpenACC turned on - replaced complicated string conditionals with case construct to improve performance and clarity - the conditionals above were moved to diagnostics init - cleaned up the thickness edge routine a little | 01 April 2022, 22:59:08 UTC |
ef86649 | Wuyin Lin | 01 April 2022, 20:38:24 UTC | Merge branch 'brhillman/atm/add-arcticx4v1-rrm' (PR #4758) Add pan-arctic atmosphere RRM grid Add a new pan-arctic atmosphere RRM grid (arcticx4v1pg2). Configuration and input files provided to run with four different configurations: arcticx4v1pg2_oARRM60to10 arcticx4v1pg2_r05_oARRM60to10 arcticx4v1pg2_r125_oARRM60to10 [BFB] | 01 April 2022, 20:38:24 UTC |
c41fd42 | Jon Wolfe | 01 April 2022, 18:22:15 UTC | Merge branch 'xylar/mpas-framework/fix-gnu-nersc' (PR #4854) MPAS standalone: Add gnu >=10 support on Cori This merge adds the -fallow-argument-mismatch flag to the gnu-nersc target for MPAS standalone builds. This got missed in #4821. [BFB] | 01 April 2022, 18:22:15 UTC |
6bf5b25 | Jon Wolfe | 01 April 2022, 18:16:22 UTC | Merge branch 'jonbob/scripts/update-anvil-lowres-layouts' (PR #4851) Update default anvil layouts for S and M sizes of ne30pg2_EC30to60E2r2 One of our users, @alicebarthel, was doing a series of ne30pg2_EC30to60E2r2 runs on anvil and the PACE output showed the S and M default layouts could be further optimized. This PR updates these default layouts with some improvement without changing the overall processor counts. S goes from ~5.0 to 5.4 sypd, and M improves from ~8.8 to 9.4 sypd. [BFB] | 01 April 2022, 18:16:22 UTC |
b0b3f16 | Jon Wolfe | 31 March 2022, 23:33:25 UTC | Update bld files with output from scripts, to match Registry changes | 31 March 2022, 23:33:25 UTC |
f04c320 | Jon Wolfe | 31 March 2022, 14:49:04 UTC | Merge branch 'cbegeman/ocn/add-explicit-layerThickEdge-vars' (PR #4832) Add explicit layerThicknessEdge variables This PR replaces the existing layerThicknessEdge variable with two variables: * layerThicknessEdgeMean refers to the mean of the neighboring cell-centered layerThickness * layerThicknessEdgeFlux refers to the layerThicknessEdge used to flux momentum and tracers. When config_thickness_flux_type = 'centered' layerThicknessEdgeFlux = layerThicknessEdgeMean, but when config_thickness_flux_type = 'upwind' layerThicknessEdgeFlux is upwinded. Prior to this PR, when config_thickness_flux_type = 'upwind' layerThicknessEdge was upwinded everywhere. This PR uses layerThicknessEdgeFlux for horizontal advection terms and layerThicknessEdgeMean elsewhere, such as for mixing parameterizations. This change is in preparation for adding additional options for thickness fluxes. This PR also removes layerThicknessEdge from global stats, since it did not offer much information not given by the global stats for layerThickness. [BFB] | 31 March 2022, 14:49:04 UTC |
0b07852 | Jon Wolfe | 31 March 2022, 14:39:54 UTC | Merge branch 'njeffery/seaice/correct_flux_script_error' (PR #4856) Fixes subscript error in ice to ocean DON carbon flux Debug mode caught a subscript error in ice-ocean BGC flux in the mpas-seaice driver and this fixes it. Only BGC results are impacted so it's BFB in physics only runs. Fixes #4855 [non-BFB] in ocean-ice bgc runs | 31 March 2022, 14:39:54 UTC |
35233fc | Azamat Mametjanov | 31 March 2022, 02:51:39 UTC | Update mpaso openacc files and flags | 31 March 2022, 02:51:39 UTC |
01d8a9f | Azamat Mametjanov | 30 March 2022, 03:05:41 UTC | Update PGI compiler flags | 30 March 2022, 03:05:41 UTC |
b5801d3 | Jon Wolfe | 08 March 2022, 16:28:16 UTC | Modify seq_diag_atm_mct to use lfrac or lfrin based on samegrid_al | 29 March 2022, 19:43:44 UTC |
db45e6b | Jon Wolfe | 08 March 2022, 16:17:36 UTC | Clean up and fix documentation strings to match the set lnd fraction | 29 March 2022, 19:41:29 UTC |
a34f043 | Jon Wolfe | 07 March 2022, 22:48:51 UTC | Make lnd fraction dependent on samegrid_al Conflicts: driver-mct/main/seq_diagBGC_mct.F90 | 29 March 2022, 19:40:18 UTC |
859daf0 | Jon Wolfe | 28 February 2022, 20:07:09 UTC | Change default handling of lnd2atm fluxes when on different grids | 29 March 2022, 19:28:28 UTC |
046869a | Jon Wolfe | 28 February 2022, 20:04:26 UTC | Fix attribute names for Fall_fco2_lnd flux field | 29 March 2022, 19:25:34 UTC |
bdd32c2 | Azamat Mametjanov | 29 March 2022, 03:33:23 UTC | Add MPI process stride for tasks exclusively owned by a component | 29 March 2022, 03:33:23 UTC |
3d43fde | jayeshkrishna | 28 March 2022, 17:12:51 UTC | Avoiding -ve (& 0) idx into the MeanMonthFlow array Since the Fortran standard does not support short circuit evaluation like C avoiding indexing the array when the index can be <= 0. Note: this issue only shows up with certain compilers + options (e.g. gfortran in debug mode) | 28 March 2022, 17:12:51 UTC |
64135f8 | Nicole Jeffery | 25 March 2022, 19:46:03 UTC | Fixes subscript error in ice to ocean DON carbon flux Corrects issue #4855 BFB in physics only runs. Non-BFB in ocean-ice bgc runs. | 25 March 2022, 19:46:03 UTC |