95af505 | Adrian K. Turner | 09 November 2021, 23:12:24 UTC | Fixed seaice output stream bug Corrected output stream name from 'output' to 'none' for several analysis members | 09 November 2021, 23:12:24 UTC |
417b03f | Jon Wolfe | 03 November 2021, 14:42:40 UTC | Merge branch 'sbrus89/ocn/improve-harmonic-analysis-options' (PR #4586) Modify harmonic analysis AM namelist options and output stream The start and end dates for harmonic analysis period have been replaced by a config_AM_harmonicAnalysis_start_delay and a config_AM_harmonicAnalysis_duration option (with units of days). This makes it easier to change the start/end date of the simulation without having to change harmonic analysis AM options. Stream output is now controlled by the AM compute routine. The HA output stream is written following the least squares solve. The output_interval in the streams file should be set for a long time interval (i.e. 1000 years) so it is not written by the AM driver routine. This means the output_interval no longer needs to be changed when the simulation length or harmonic analysis period changes. [NML] [BFB] | 03 November 2021, 14:42:40 UTC |
3a87fa1 | Balwinder Singh | 01 November 2021, 20:28:05 UTC | Merge branch 'brhillman/mmf/integrate-rrtmgpxx-new' (PR #4585) Enable using the C++ back end of RRTMGP for both E3SM and E3SM-MMF. The C++ code uses YAKL to enable running on CPU or GPU with the same codebase. Maintains the ability to run the F90 back end as well (remains the default). Building with the C++ version is enabled by adding -rrtmgpxx to CAM_CONFIG_OPTS in env_build.xml. Note that this is in addition to using the -rad rrtmgp option. The two flags are required because -rad rrtmgp sets much of the logic that determines input data and settings in configure and build-namelist, which we do not want to duplicate just because we are running the C++ back end instead of the F90 back end. A testmod (eam-rrtmgpxx) is also added so that RRTMGPXX can be built via create_test without any user interaction. Also updates both the YAKL and RRTMGP submodules. Believe it or not, this is BFB. [BFB] * brhillman/mmf/integrate-rrtmgpxx-new: (71 commits) Explicitly use pool allocator for ESMT test Make sure rrtmgp_interface gets netcdf includes Only copy to/from device as needed Upgrading to new YAKL with a new pool allocator that uses the avilable space more efficiently and turns off the pool for targets without a separate memory address space. Add RRTMGPXX test to E3SM integration Make RRTMGPXX default for all MMF compsets Removed unused commented out code Revert to RRTMGP I/O Update to latest RRTMGP master after rrtmgp_const fix Add radiation_final for CRM rrtmg Remove netcdf-cxx4 from submodules Add radiation_final to rrtmg interfaces Removing unused declarations, and changing rrtmgp interface build system to use the new YAKL target-based approach Changing const.h to rrtmgp_const.h for rrtmgpxx Getting rid of fast math in DEBUG CUDA_FLAGS, and adding -g to DEBUG CUDA_FLAGS Update to latest RRTMGP master with bugfix Adding debug flags to CXXFLAGS for summit+gnu Wrapping data in host Arrays, then copying to device, then copying back at the end; for SW and LW Moving to latest YAKL, which no longer fails upon multiple calls to yakl::initialize and yakl::finalize Fix CUDA_FLAGS for DEBUG=TRUE ... | 01 November 2021, 20:28:05 UTC |
134922b | Benjamin Hillman | 29 October 2021, 21:55:33 UTC | Explicitly use pool allocator for ESMT test Make sure we use the pool allocator for the ESMT test. This used to be the default in YAKL, but commit fcf7ba19 appears to have changed that behavior, and for some reason that is causing non-BFB behavior with this test. We do not understand why, but reverting this behavior should make this test BFB with master, and get the ESMT test to pass again on Chrysalis. | 29 October 2021, 21:55:33 UTC |
d63ecda | Min Xu | 29 October 2021, 18:41:58 UTC | Merge branch 'ykim/spock/gnu' (PR #4529) Add CIME configuration for GNU compiler at ORNL Spock The following configuration files are updated with adding GNU Spock sections: config_machines.xml config_batch.xml config_pio.xml This also adds gnu_spock.cmake (supports new cmake-based build). Fixes #4151 [BFB] | 29 October 2021, 18:41:58 UTC |
3074dde | Jon Wolfe | 29 October 2021, 15:11:35 UTC | Merge branch 'jonbob/mpas-seaice/update-orrs18to6v3-initial' (PR #4621) Change extents of initial polar disks for oRRS18to6v3 grid Changes the default settings for the latitude extents for the oRRS18to6v3 grid. These settings define the initial polar seaice disks when running without spunup initial conditions, and are now consistent with those for other more resolved grids. Previously G-case configurations failed when run with default timesteps for this grid, and testing of this PR successfully completed 15 days. Fixes #4610 [non-BFB] for G-case configurations using oRRS18to6v3, but BFB for all tested configurations | 29 October 2021, 15:11:35 UTC |
dca6601 | Jon Wolfe | 29 October 2021, 15:07:07 UTC | Merge branch 'darincomeau/scripts/add-cryo-4xco2-compset' (PR #4628) Adding CRYO1850-4xCO2 compset Adds a new Cryosphere compset that is similar to WCYCL1850-4xCO2, except uses Cryosphere settings for Antarctic runoff (disables Antarctic runoff from the coupler to the ocn, includes ice-shelf melt fluxes and data icebergs). [BFB] for all currently tested configurations | 29 October 2021, 15:07:07 UTC |
2105c90 | Jon Wolfe | 28 October 2021, 15:22:10 UTC | Update bld files with output from automated scripts to match Registry | 28 October 2021, 15:22:10 UTC |
9b9a6bf | Jon Wolfe | 28 October 2021, 15:16:24 UTC | Merge branch 'jonbob/mpas-ocean/update-bld-files' (PR #4625) Update bld files to match current Registry using automated scripts Some of the intermediate bld files were not updated with formats consistent with our scripts used for autogenerating them from Registry. This PR updates them so they are. [BFB] | 28 October 2021, 15:16:24 UTC |
0999c64 | Darin Comeau | 27 October 2021, 21:28:03 UTC | Adding CRYO1850-4xCO2 compset | 27 October 2021, 21:28:03 UTC |
b807c76 | Jon Wolfe | 27 October 2021, 17:34:38 UTC | Update bld files to match current Registry using automated scripts | 27 October 2021, 17:34:38 UTC |
8cf1f79 | Benjamin Hillman | 26 October 2021, 19:17:51 UTC | Make sure rrtmgp_interface gets netcdf includes | 26 October 2021, 19:17:51 UTC |
7770b27 | Robert Jacob | 25 October 2021, 20:34:07 UTC | Merge branch 'ambrad/driver/kokkos-init-fin' (PR #4614) When multiple components use Kokkos, we need to be careful about initializing and finalizing Kokkos. Each can occur only once, and finalization must occur after the last component using Kokkos has deallocated its views. This commit modifies cime_comp_mod to finalize components in the opposite order that it initializes them. Then a component can use is_initialized to determine if it is in charge of initializing and finalizing Kokkos. In this commit, I've modified SL transport to follow this pattern. Fixes #4606. [BFB] | 25 October 2021, 20:34:07 UTC |
9aba0a2 | Youngsung Kim | 25 October 2021, 17:27:24 UTC | set sbetr to a previous commit to prevent CENT_ECACNP issue at sbetr | 25 October 2021, 17:27:24 UTC |
3d2d461 | Peter Schwartz | 25 October 2021, 16:21:07 UTC | Merge branch 'jinyuntang/lnd/bug_fix' (PR #4440) This fixed issue #4428 and issue #4423. [BFB] | 25 October 2021, 16:21:07 UTC |
e04228f | Youngsung Kim | 25 October 2021, 15:29:24 UTC | set sbetr submodule to a newer commit that fixed ODEMod.F90 issue | 25 October 2021, 15:29:24 UTC |
8e82c97 | Jon Wolfe | 25 October 2021, 15:04:53 UTC | Change extents of initial polar disks for oRRS18to6v3 grid | 25 October 2021, 15:04:53 UTC |
38c93a6 | Youngsung Kim | 25 October 2021, 14:42:27 UTC | Merge branch 'master' into ykim/spock/gnu | 25 October 2021, 14:42:27 UTC |
286db1e | Youngsung Kim | 25 October 2021, 14:41:22 UTC | created gnu_spock.cmake to support the new cmake build system | 25 October 2021, 14:41:22 UTC |
bcde284 | Peter Schwartz | 22 October 2021, 19:19:25 UTC | Merge branch 'bishtgautam/lnd/ssp3-rcp7'(PR #4601) The namelist file is updated to include LULC files for SSP3 RCP7.0 [BFB] | 22 October 2021, 19:19:25 UTC |
475a9e3 | Benjamin Hillman | 22 October 2021, 18:00:43 UTC | Only copy to/from device as needed | 22 October 2021, 18:00:43 UTC |
ca2a217 | Jon Wolfe | 22 October 2021, 17:30:26 UTC | Merge branch 'philipwjones/ocn/eos-accfix' (PR #4578) Switched from device_resident to create for small arrays in ocn EOS The nvhpc 21.3 compiler on Summit was failing to build the OpenACC form of the ocean equation_of_state_jm routine with the device_resident declaration for small arrays. This fix switches to the more generic OpenACC create/delete paradigm for these arrays while still retaining device_resident for the larger T,S temporaries. Change is bit-for-bit for both CPU and GPU configurations (impacts data transfers only) as tested on a QU240 case with the JM EOS enabled. Fixes #4577 [BFB] | 22 October 2021, 17:30:26 UTC |
8d9152f | Steven Brus | 22 October 2021, 16:28:36 UTC | Explain the HA start date/length are relative to original start date | 22 October 2021, 16:28:36 UTC |
230e8d8 | Steven Brus | 22 October 2021, 15:54:21 UTC | Add forceWriteNow to HA AM stream write call | 22 October 2021, 15:54:21 UTC |
efe1cb1 | Peter Schwartz | 21 October 2021, 20:38:10 UTC | Merge branch 'bishtgautam/lnd/updates-crop-test' (PR #4572) Uses updated netcdf files that are in netcdf-3 format Fixes #4570 [BFB] | 21 October 2021, 20:38:10 UTC |
4131c6e | Jon Wolfe | 21 October 2021, 19:00:02 UTC | Merge branch 'maltrud/ocean/add-marbl-bgc' (PR #4587) Add MARBL ocean BGC library This PR adds the MARBL ocean Biogeochemistry library to MPAS-Ocean. this replaces the current BGC routines which were a direct port of the BEC routines from CESM1, so the E3SMv1 implementation will no longer be supported. this will allow E3SM to stay more up to date with BGC improvements made by the community. in E3SM, MARBL is a submodule which was forked from the public MARBL repository and set to the CESM2.1 tag. there are several important notes about this PR: * ocean BGC will not run “out of the box” since a new MARBL specific initial condition file is required. this will be added to the config system once full validation has been finished. * MARBL is not yet compatible with sea ice BGC due to the addition of refractory DOC. updates to the interface with sea ice BGC will happen soon. * MARBL does not simulate Phaeocyctis as a phytoplankton functional type. there is an unsolved issue with the ALT_CO2 tracer that can cause MARBL to post error warnings. this tracer is essentially a diagnostic so does not affect the simulation. this will be sleuthed out soon. * testing has been done using the EC30to60E2r2 grid. comparisons with the v1 BGC after 10 simulated years (using identical forcing) show no obvious problems. exact restart and independence of solution with respect to processor layout have been verified on anvil with this setup. * it isn't clear yet how to interface with MARBL within an ice shelf cavity, so this PR does not support land ice grids. [non-BFB] for BGC configurations only | 21 October 2021, 19:00:02 UTC |
1bc9501 | Youngsung Kim | 21 October 2021, 14:17:19 UTC | updated sbetr to the commmit that updates ODEMod.F90 | 21 October 2021, 14:17:19 UTC |
1250a77 | Youngsung Kim | 21 October 2021, 14:14:57 UTC | resolved merge conflict at cime_config/machines/config_compilers.xml | 21 October 2021, 14:14:57 UTC |
44532f5 | Youngsung Kim | 21 October 2021, 13:50:43 UTC | Merge branch 'master' into ykim/spock/gnu | 21 October 2021, 13:50:43 UTC |
c3bb914 | Youngsung Kim | 21 October 2021, 13:49:55 UTC | on spock machine, cray compiler wrappers are used instead of native gnu compiler | 21 October 2021, 13:49:55 UTC |
d3f2973 | Jon Wolfe | 20 October 2021, 22:15:25 UTC | Temorarily remove ocean BGC from tested BGCEXP configurations | 20 October 2021, 22:15:25 UTC |
d8e768e | James Foucar | 20 October 2021, 21:28:36 UTC | Merge branch 'jgfouca/clean_and_fix_macros' into master (PR #4611) Clean and fix cmake macros One surprise from the legacy system is that: <FIELD> <base> stuff </base> <append> more-stuff </append> </FIELD> in the above XML, the <base> element acts just like an <append> if FIELD has already been defined in one of the parent blocks. The converter did not take this into account and so some flags/settings were being lost in the system. This PR fixes the issue by always setting list variables via string(APPEND ...) and never set(...). Also, remove configs for some unused OSs. Also, restore some changes to anlworkstation that I believe were unintentional Also, begin work on a new evaluator script to help clean and sanitize cmake macros. Fixes #4595 [BFB] * jgfouca/clean_and_fix_macros: Begin work on new evaluator script Restore SZIP_PATH and ZLIB_PATH to anlworkstation Remove configs for unused OSes | 20 October 2021, 21:28:36 UTC |
92b553d | jayeshkrishna | 20 October 2021, 16:33:12 UTC | Merge branch 'dqwu/anlmachines/add-perl5lib' (PR #4612) Some E3SM cases use the Switch module of Perl. However, on ANL machines, Switch.pm is not available in the default Perl paths. Setting env var to point to the shared local Perl lib (contains Switch.pm) installed in the climate directory on ANL machines. On anlworkstation and anlgce, E3SM users no longer need to install a local Perl lib in their home directories and use it to locate Switch.pm. [BFB] | 20 October 2021, 16:33:12 UTC |
818546b | Matt Norman | 19 October 2021, 19:42:46 UTC | Upgrading to new YAKL with a new pool allocator that uses the avilable space more efficiently and turns off the pool for targets without a separate memory address space. | 19 October 2021, 19:42:46 UTC |
419db29 | dqwu | 19 October 2021, 15:00:19 UTC | Setting PERL5LIB to a shared local Perl lib for ANL machines Some E3SM cases use the Switch module of Perl. However, on ANL machines, Switch.pm is not available in the default Perl paths. Setting env var to point to the shared local Perl lib (contains Switch.pm) installed in the climate directory on ANL machines. | 19 October 2021, 16:28:40 UTC |
ac5e9da | Steven Brus | 19 October 2021, 15:06:33 UTC | Revise harmonicAnalysisOutput stream parameters | 19 October 2021, 15:06:33 UTC |
b936852 | Wuyin Lin | 19 October 2021, 14:52:47 UTC | Merge branch '/tangq/lnd/NARRM-historical' (PR #4603) Add the default land use time series file for NA RRM The land use time series setting is missing for the North America (NA) regionally refined model settings. This PR adds the default file for the DECK historical runs. [BFB] only impact historical simulation with NARRM [NML] specify the flanduse_timeseries file. | 19 October 2021, 14:52:47 UTC |
d730e06 | Jon Wolfe | 19 October 2021, 14:39:50 UTC | Merge branch 'vanroekel/ocean/fix-conservation-check-restart-stream' (PR #4605) Fixes typo in ocean conservationCheckRestart There is a small typo in the output stream for conservationCheckRestart, where the file template is mpassi, but should be mpaso. Fixes #4604 [BFB] | 19 October 2021, 14:39:50 UTC |
af89db7 | James Foucar | 18 October 2021, 22:44:25 UTC | Begin work on new evaluator script Should be useful for checking, cleaning, and sanitizing our macros | 18 October 2021, 22:44:25 UTC |
e2e02e2 | James Foucar | 18 October 2021, 22:42:22 UTC | Restore SZIP_PATH and ZLIB_PATH to anlworkstation I believe these were removed unintentionally Change operations on list variables to always use STRING(APPEND ...). Never set() these variables, it's too easy to lose previous settings and this isn't semantically compatible with the legacy system. | 18 October 2021, 22:42:25 UTC |
27207d9 | James Foucar | 18 October 2021, 20:43:12 UTC | Remove configs for unused OSes | 18 October 2021, 20:43:12 UTC |
9a2742e | Peter Schwartz | 18 October 2021, 20:38:04 UTC | Merge remote-tracking branch 'rgknoxfork/rgknox/lnd/fates-cnp' (PR #4325) This set of changes finalizes the first version of full coupling of Nitrogen, Phosphorus and methane between ELM and FATES. Nitrogen (in the form of NH4 and NO3) and Phosphorus are removed from mineralized soil pools in competition with soil actors for uptake by plants (ELM->FATES). Upon turnover, plants will pass N and P back to the litter pool (FATES->ELM) where it is decomposed and re-enters the soil ecosystem. While methane (CH4) is not "passed" between the two models, FATES sends information about rooting profiles and NPP to allow the methane module to function, which enables the functioning of the nitirification and denitrification processes. In summary, these changes enabled compatibility with FATES API 16.0.0. [non-BFB] only for FATES and ECACNTBC compsets. | 18 October 2021, 20:38:04 UTC |
1ae1b3c | Andrew M. Bradley | 18 October 2021, 20:00:17 UTC | Improve Kokkos initialization and finalization. When multiple components use Kokkos, we need to be careful about initializing and finalizing Kokkos. Each can occur only once, and finalization must occur after the last component using Kokkos has deallocated its views. This commit modifies cime_comp_mod to finalize components in the opposite order that it initializes them. Then a component can use is_initialized to determine if it is in charge of initializing and finalizing Kokkos. In this commit, I've modified SL transport to follow this pattern. | 18 October 2021, 20:00:17 UTC |
0202874 | Ryan Knox | 18 October 2021, 18:47:28 UTC | Updating fates submodule pointer to include ibm compiler fixes,tag: sci.1.46.1_api.16.0.0_IBMFix. | 18 October 2021, 18:47:28 UTC |
367617c | Luke Van Roekel | 17 October 2021, 04:58:16 UTC | Fixes typo in ocean conservationCheckRestart | 17 October 2021, 04:58:16 UTC |
60fb8f6 | Qi Tang | 16 October 2021, 04:13:35 UTC | Add the default land use time series file for NA RRM | 16 October 2021, 04:13:35 UTC |
55d0464 | noel | 15 October 2021, 21:29:11 UTC | Merge branch E3SM-Project/azamat/benching/update-v2-grids-compsets (PR #4551) Update V2 benchmarking grids and compsets | 15 October 2021, 21:29:11 UTC |
38fb21b | Gautam Bisht | 15 October 2021, 20:10:36 UTC | Adds LULC files for SSP3 RCP 7.0 The namelist file is updated to include LULC files for SSP3 RCP7.0 [BFB] | 15 October 2021, 20:21:14 UTC |
e008fa1 | James Foucar | 14 October 2021, 17:46:50 UTC | Merge branch 'jgfouca/cime_update_2021_10_11' into master (PR #4597) CIME Update to a26c8e393d4e062459b6dfb9605700fe24665e50 Changes 1) Remove MODEL as a setting for the build system (Use COMP_NAME) 2) remove compare vars dependency in cprnc 3) extend the recent cmake macro changes to allow the inclusion of CMAKE_LANG_COMPILER_VERSION in cmake_macros files (Should no longer need intel17, intel18 compiler names!) 4) cime_bisect: Improve robustness when using custom script Fixes: 1) Using custom cmake macros in ~/.cime 2) Fix typo in scripts/Tools/Makefile [BFB] * jgfouca/cime_update_2021_10_11: Grab a couple more things from CIME that we will want CIME Update | 14 October 2021, 17:46:50 UTC |
ce24b52 | jayeshkrishna | 14 October 2021, 15:48:46 UTC | Merge branch 'dqwu/machinefiles/anlgce' (PR #4598) This PR allows machine anlgce to run all e3sm_developer tests. [BFB] | 14 October 2021, 15:48:46 UTC |
b995634 | dqwu | 13 October 2021, 17:22:06 UTC | Changing max tasks per node to 32 (from 16) for anlgce This setting (also used by anlworkstation) allows some E3SM tests like SMS_P12x2.* to run on anlgce. | 13 October 2021, 17:22:06 UTC |
6b8a9e1 | Mark Petersen | 12 October 2021, 22:23:23 UTC | Clean up white space | 12 October 2021, 22:23:23 UTC |
1ecee87 | Mark Petersen | 12 October 2021, 22:14:19 UTC | Turn verbose false to prevent print to screen | 12 October 2021, 22:14:19 UTC |
ee5eb80 | Mark Petersen | 12 October 2021, 15:25:39 UTC | update marbl submodule | 12 October 2021, 22:12:38 UTC |
ad04d54 | James Foucar | 12 October 2021, 21:58:20 UTC | Grab a couple more things from CIME that we will want | 12 October 2021, 21:58:20 UTC |
9f8cdbf | Benjamin Hillman | 12 October 2021, 20:24:33 UTC | Add RRTMGPXX test to E3SM integration | 12 October 2021, 20:24:33 UTC |
7a84677 | Benjamin Hillman | 12 October 2021, 20:22:50 UTC | Make RRTMGPXX default for all MMF compsets | 12 October 2021, 20:22:50 UTC |
aa53caf | jayeshkrishna | 12 October 2021, 17:57:54 UTC | Merge branch 'jayeshkrishna/machinefiles/anlgce' (PR #4563) Adding support for the ANL CELS GCE compute nodes [BFB] | 12 October 2021, 17:57:54 UTC |
e61ea3c | James Foucar | 12 October 2021, 15:34:39 UTC | CIME Update to 4a6b6043888591ab5ecb41360864014eda5862f4 Changes 1) Remove MODEL as a setting for the build system (Use COMP_NAME) 2) remove compare vars dependency in cprnc 3) extend the recent cmake macro changes to allow the inclusion of CMAKE_LANG_COMPILER_VERSION in cmake_macros files - (Should no longer need intel17, intel18 compiler names!) 4) cime_bisect: Improve robustness when using custom script [BFB] | 12 October 2021, 15:34:39 UTC |
f7b2f3f | Ryan Knox | 12 October 2021, 14:49:52 UTC | Enclosed a deallocation in logic that only deallocates when ECA/MIC is active, as the memory is only allocated when it is active. | 12 October 2021, 14:49:52 UTC |
17a3044 | Azamat Mametjanov | 11 October 2021, 22:34:15 UTC | Merge branch 'azamat/chrysalis/add-lores-f-compset-pelayouts' (PR #4592) Add ne30pg2_EC30to60E2r2 F-compset PE-layouts on Chrysalis Fixes E3SM-Project/E3SM#4588 [BFB] | 11 October 2021, 22:34:15 UTC |
62ffe68 | Phil Jones | 11 October 2021, 20:10:46 UTC | fixes MARBL paths and build arguments for ocn standalone build fixes the include and build paths for the MARBL library sends E3SM build arguments to the MARBL Makefile requires a companion fix to the MARBL library Makefile | 11 October 2021, 20:10:46 UTC |
e17cc6a | James Foucar | 11 October 2021, 16:14:11 UTC | Merge branch 'jgfouca/add_macro_cmakelists' into master (PR #4593) Add CMakeLists.txt to cmake_macros This will be used in the next CIME update for the latest build system features (support for version checking in macros). [BFB] * jgfouca/add_macro_cmakelists: Add CMakeLists.txt to cmake_macros | 11 October 2021, 16:14:11 UTC |
ab9786b | James Foucar | 11 October 2021, 16:12:16 UTC | Add CMakeLists.txt to cmake_macros This will be used in the next CIME update for the latest build system features (support for version checking in macros). | 11 October 2021, 16:12:16 UTC |
9426b05 | Azamat Mametjanov | 09 October 2021, 00:10:56 UTC | Add low-res F-compset PE-layouts on Chrysalis | 09 October 2021, 00:10:56 UTC |
a02cd55 | Benjamin R. Hillman | 08 October 2021, 20:15:19 UTC | Removed unused commented out code | 08 October 2021, 20:15:19 UTC |
520de2b | jayeshkrishna | 08 October 2021, 16:45:32 UTC | Adding cmake macro for anlgce Adding the cmake macros file for ANL CELS GCE compute nodes. The macro file was generated using the <E3SM_SRC>/cime_config/machines/scripts/converter script. | 08 October 2021, 16:45:32 UTC |
6c74a42 | jayeshkrishna | 07 October 2021, 16:38:33 UTC | Force PE layout with 16 procs on anlgce Using the same PE layout as anlworkstation (old ANL compute nodes) for the new ANL CELS compute nodes | 08 October 2021, 16:15:36 UTC |
1004cd2 | jayeshkrishna | 22 September 2021, 15:56:30 UTC | Adding support for ANL GCE nodes Adding support for the ANL CELS GCE compute nodes | 08 October 2021, 16:15:36 UTC |
2dd516e | Jon Wolfe | 08 October 2021, 14:42:50 UTC | Merge branch 'philipwjones/ocn/del4fix' (PR #4583) Fix allocate error in ocean del4 tracer mixing Fixes a bug in which a temp was being allocated before one of the dimensions was defined, causing a malloc corruption error Because this was a pretty egregious error, it's unlikely any of our current tests actually exercise this code and I could only verify it runs to completion and gives bit-for-bit when del4 is off. Fixes #4581 [BFB] | 08 October 2021, 14:42:50 UTC |
e77b3dd | Benjamin R. Hillman | 07 October 2021, 20:47:29 UTC | Revert to RRTMGP I/O | 07 October 2021, 20:47:29 UTC |
a033221 | Mathew Maltrud | 07 October 2021, 19:59:01 UTC | remove ecosys_diag_auto_graze_TOT from streams.ocean | 07 October 2021, 19:59:01 UTC |
4c32474 | James Foucar | 07 October 2021, 19:49:14 UTC | Merge branch 'jgfouca/replace_xml_with_cmake_macros' into master (PR #4537) Major build system enhancement: deprecate config_compilers.xml! This is the most significant change to the case-control system since the cmake-ification (#3043) effort 2 years ago. This PR is a natural follow-on to that effort that further cmake-ifies our build system and reduces technical debt / complexity in the build system by stripping away layers of CIME magic that were not adding much value for us. Specifically, the config_compilers.xml file, which CIME uses to generate the Macros.make and Macros.cmake casedir files (specifies compilers, flags, link flags etc), will be replaced with hierarchical CMake cache files that are persistently stored in our repo. [BFB] * jgfouca/replace_xml_with_cmake_macros: Fix handling of '\!' Add CIME hotfix Use COMP_NAME, not MODEL Sync with latest config_compilers.xml Add support for compiler_os.cmake macros Improve dox for compare-flags compare-flags: make tool more flexible and better output Refresh cache files after upstream merge Add some helper scripts for this effort Update CIME with latest Add support for universal macro Update CIME to bring in new macro capability Remove obsolete comment Add new cmake-based macros | 07 October 2021, 19:49:14 UTC |
62cbc03 | Benjamin R. Hillman | 07 October 2021, 19:35:55 UTC | Update to latest RRTMGP master after rrtmgp_const fix | 07 October 2021, 19:35:55 UTC |
131366c | Benjamin R. Hillman | 07 October 2021, 19:35:25 UTC | Add radiation_final for CRM rrtmg | 07 October 2021, 19:35:25 UTC |
6819002 | Steven Brus | 07 October 2021, 15:34:24 UTC | Modify harmonic analysis AM namelist options and output stream - Start and end dates for the harmonic analysis period have been replaced by a start_delay and a duration option (with units of days) - Stream output is now controlled by the AM compute routine. The HA output stream is written following the least squares solve. The output_interval in the streams file should be set for a long time interval (i.e. 1000 years) so it is not written by the AM driver routine. | 07 October 2021, 16:48:45 UTC |
347a118 | Benjamin R. Hillman | 07 October 2021, 15:45:53 UTC | Remove netcdf-cxx4 from submodules | 07 October 2021, 15:45:53 UTC |
b18a910 | Benjamin R. Hillman | 07 October 2021, 15:42:11 UTC | Add radiation_final to rrtmg interfaces | 07 October 2021, 15:42:11 UTC |
7628ed0 | Jon Wolfe | 07 October 2021, 14:37:58 UTC | Merge branch 'jonbob/mpaso/fix-xml-cfc' (PR #4580) Fix quotes to agree with XML format Fixes a minor XML issue in the Registry_CFC.xml file in mpas-ocean. This will allow the XML parser we use to autogenerate bld files to work. Fixes #4579 [BFB] | 07 October 2021, 14:37:58 UTC |
07ff719 | Gautam Bisht | 07 October 2021, 14:02:23 UTC | Merge branch 'tktesfa2014/lnd/archv2_disag_grd_to_topo_rebase2' (PR #4273) Implementation of topography-based subgrid structure (topounits) and downscaling methods of atmospheric forcing from atmosphere grid to land topounits. It allows variable number of topounits per grid improving computational demand as compared to having the same number of topounits per grid for all grids. Downscaling methods include two approaches of precipitation downscaling developed at PNNL. Also, this branch includes downscaling methods for other atmospheric forcing developed at the University of Arizona. In addition, topounits have been implemented into the dynamic subgrid code and data structures of recent E3SM features such as soil erosion, irrigation and plant hydraulics. [BFB] | 07 October 2021, 14:02:23 UTC |
4b3a034 | Matt Norman | 07 October 2021, 13:49:07 UTC | Removing unused declarations, and changing rrtmgp interface build system to use the new YAKL target-based approach | 07 October 2021, 13:49:07 UTC |
92413e0 | Matt Norman | 07 October 2021, 13:17:10 UTC | Changing const.h to rrtmgp_const.h for rrtmgpxx | 07 October 2021, 13:17:10 UTC |
774e552 | Matt Norman | 07 October 2021, 13:02:32 UTC | Getting rid of fast math in DEBUG CUDA_FLAGS, and adding -g to DEBUG CUDA_FLAGS | 07 October 2021, 13:02:32 UTC |
979ba18 | Mathew Maltrud | 06 October 2021, 22:28:24 UTC | fix mem leak by erasing marbl log every timestep | 06 October 2021, 22:28:24 UTC |
1d27313 | Benjamin Hillman | 06 October 2021, 16:10:44 UTC | Update to latest RRTMGP master with bugfix | 06 October 2021, 18:53:48 UTC |
ad0f95a | Matt Norman | 06 October 2021, 15:10:14 UTC | Adding debug flags to CXXFLAGS for summit+gnu | 06 October 2021, 18:53:48 UTC |
52c8c7b | Matt Norman | 05 October 2021, 19:58:47 UTC | Wrapping data in host Arrays, then copying to device, then copying back at the end; for SW and LW | 06 October 2021, 18:53:36 UTC |
8b0d610 | Matt Norman | 05 October 2021, 14:21:04 UTC | Moving to latest YAKL, which no longer fails upon multiple calls to yakl::initialize and yakl::finalize | 06 October 2021, 18:53:36 UTC |
1ef3201 | Benjamin Hillman | 04 October 2021, 20:09:45 UTC | Fix CUDA_FLAGS for DEBUG=TRUE Adding -g to the CUDA_FLAGS in DEBUG=TRUE mode. Current stack trace for MMF rrtmgpxx on the GPU is a segfault during finalize with -DYAKL_DEBUG. | 06 October 2021, 18:53:21 UTC |
bea6c91 | Benjamin Hillman | 24 September 2021, 20:09:37 UTC | Fix some class references in kernels | 06 October 2021, 18:52:59 UTC |
603c5d1 | Matt Norman | 10 September 2021, 22:19:42 UTC | Updating scripts for summit and a bit of the standalone build system | 06 October 2021, 18:52:59 UTC |
59a9c36 | Benjamin Hillman | 05 October 2021, 04:29:39 UTC | Fix flux_avg for debug mode (again) | 06 October 2021, 18:52:59 UTC |
b516ae8 | Benjamin Hillman | 09 September 2021, 22:49:30 UTC | Add a testmod for RRTMGPXX | 06 October 2021, 18:52:59 UTC |
4b8f3aa | Benjamin Hillman | 27 August 2021, 19:47:37 UTC | Fixes for new YAKL updates | 06 October 2021, 18:52:50 UTC |
df9c3ce | Benjamin Hillman | 26 August 2021, 04:12:16 UTC | Fix cmake build for RRTMGPXX | 06 October 2021, 18:52:30 UTC |
1a44107 | Benjamin Hillman | 26 August 2021, 04:11:12 UTC | Need to explicitly copy fluxes for GPU | 06 October 2021, 18:51:44 UTC |
7931a7f | Benjamin Hillman | 13 August 2021, 16:25:48 UTC | Add dim args for RRTMGPXX changes | 06 October 2021, 18:51:44 UTC |
5c4b9c7 | Benjamin Hillman | 02 June 2021, 23:05:28 UTC | Fix RRTMGPXX build | 06 October 2021, 18:51:27 UTC |
4e7856c | Benjamin R. Hillman | 23 March 2021, 23:08:38 UTC | Make MMF interface consistent with E3SM | 06 October 2021, 18:44:56 UTC |
9ed7a37 | Benjamin R. Hillman | 18 March 2021, 23:05:52 UTC | Make C++ and F90 versions compatible | 06 October 2021, 18:44:48 UTC |