Skip to content

Releases: GEOS-ESM/MAPL

MAPL 2.1 Initial Release

17 Apr 12:39
e23f20a
Compare
Choose a tag to compare

This is a minor release of MAPL. All testing so far shows it to be zero-diff when running the GEOS GCM. However, it does contain a number of updates and fixes to MAPL 2.0.

NOTE: To use this version of MAPL requires a Baselibs of 6.0.10 or higher. It now requires gFTL, gFTL-shared, yaFyaml, and pFlogger.

Changed

  • Corrected handling of Equation of Time in orbit (off by default)
  • Made ASSERT in ExtData more explicit in case of missing variables.
  • (re) Introduced MAPL Profiling package
  • Improved diagnostic message in HistoryGridComp for misspelled fields/bundles
  • Removed CVS keywords

Fixed

  • Corrected Python code generator scripts for component import/export specs.
  • Add directories to .gitignore for building with mepo
  • Bug building with mixed Intel/GCC compilers
  • Implemented workaround to cmake error that happens when building tests in parallel.
  • Set correct ESMA_env tag in components.yaml
  • Updated components.yaml to be inline with GEOSgcm
  • Minor problem in GMAO_pFIO Cmakelists (consistency with PRIVATE)

Removed

  • Removed support for checkout_externals and moved solely to mepo
    • Removed Externals.cfg
    • Removed checkout_externals code in CMakeLists.txt

Added

  • Added configuration for CircleCI and Github Actions
    • Builds MAPL using GCC 9.2.0 and Open MPI 4.0.2
    • Builds and runs pFIO_tests and MAPL_Base_tests
  • Add precession of equinox (not on by default)
  • Imported Python/MAPL subdir (old, but never imported to GitHub)
  • Python automatic code generator for grid comp include files
  • Added support to use pFlogger for logging
    • Command line option: --logging_config=

Fix to suppress MPI_Finalize errors

15 Apr 17:00
8dd1cd4
Compare
Choose a tag to compare

Added code to suppress (seemingly) spurious MPI_Finalize errors at end of model run. Suppression does not happen if built with -DCMAKE_BUILD_TYPE=Debug

Fix for MPI Tag

13 Apr 12:54
a4fc845
Compare
Choose a tag to compare

This release fixes an issue with a "too-large" MPI tag that presents on SLES 12 at NCCS.

History locstream bug fix

03 Apr 16:26
c2f906d
Compare
Choose a tag to compare

Fixes an issue when regridding thru the locstream in the history component.

Fix logic bug in Profilier

19 Mar 15:42
8fe80c2
Compare
Choose a tag to compare

This fixes a logic bug in the MAPL Profilers that make affect certain runs when using NUOPC.

Fix for ExtData with Low-Res inputs at high-resolutions

10 Mar 13:26
0e9cc3f
Compare
Choose a tag to compare

This release contains a fix for reading low-resolution ExtData inputs when running a high-resolution model.

Fix for Tripolar Grid

02 Mar 16:10
4b99ddd
Compare
Choose a tag to compare

This release restores tripolar grid functionality that got lost during the MAPL 2 transition.

MAPL 2.0 Release

07 Feb 17:40
c3bc53f
Compare
Choose a tag to compare

This is a major release of MAPL that is non-zero diff to MAPL v1. Changes include:

Added

  • New IO server implemented in PFIO library.
  • New command line arguments to the MAPL_Cap to run multiple input and output servers on dedicated resources.

Changed

  • Code that uses MAPL should now use MAPL instead of use MAPL_Mod.
  • CMakeLists.txt using MAPL should now have dependencies to MAPL and not MAPL_Base.
  • History and ExtData component use the PFIO IO server for all file access. Default mode is to run the IO servers on the same resources as the application.
  • The ExtData and History components use ESMF regridding for all operations and replace the FV3 regridding routines used for bilinear regridding and the MAPL tiling regridder for conservative regridding.

Vector read bug fix

09 Dec 19:13
d3c5c25
Compare
Choose a tag to compare

Bug fix for vectors read in MAPL_CFIO

Various Bugfixes

04 Dec 18:19
22a130f
Compare
Choose a tag to compare

Added

  • Added CHANGELOG.md

Fixed

  • Check status of round robin and make sure that the nodearray is allocated
  • Allow per-cell counters to be properly reset (if they are needed)
  • Must create file unit on all processors (all_pes=.true.) when writing binary History output