Update for better vectorization with gcc #1793
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
TYPE: new feature KEYWORDS: WSM6, TL/AD, 4DVAR Description of Change: This PR adds a regularized version (i.e., change discontinuous functions to continuous functions) of WSM6 microphysics scheme (MP option 106, named as 'WSM6R') and its tangent linear and adjoint (TL/AD), which enables WRF-4DVar to run with ice-phase hydrometeor analysis variables. Note that the non-linear version of WSM6R is kept just for the code reference of deriving TL/AD, and it is NOT recommended to be used for the WRF model forecast. In addition, there is an improved handling of the file unit related to the background error covariance. SOURCE: Sen YANG, Deqin LI, Liqiang CHEN (Institute of Atmospheric Environment, China Meteorological Administration, Shenyang) LIST OF MODIFIED FILES: M Registry/Registry.EM_COMMON M Registry/registry.var M Registry/registry.wrfplus M main/depend.common M phys/Makefile M phys/module_microphysics_driver.F A phys/module_mp_wsm6s.F M var/da/da_setup_structures/da_scale_background_errors.inc M var/da/da_setup_structures/da_setup_be_regional.inc M var/da/da_transfer_model/da_transfer_wrftltoxa.inc M var/da/da_transfer_model/da_transfer_wrftltoxa_adj.inc M var/da/da_transfer_model/da_transfer_xatowrftl.inc M var/da/da_transfer_model/da_transfer_xatowrftl_adj.inc M wrftladj/Makefile M wrftladj/depend.wrftladj M wrftladj/module_microphysics_driver_ad.F M wrftladj/module_microphysics_driver_tl.F A wrftladj/module_mp_wsm6s_ad.F A wrftladj/module_mp_wsm6s_tl.F M wrftladj/solve_em_ad.F M wrftladj/solve_em_tl.F TESTS CONDUCTED: 1. Jenkins tests all passed; 2. WRFDA regression test passed on Cheyenne; 3. wrfplus and 4dvar tests succeeded using mp_physics & mp_physics_ad=106. RELEASE NOTE: Add a regularized version of WSM6 and its TL/AD for 4DVar with ice-phase hydrometeor analysis variables. Yang, S., D. Q. Li, L. Q. Chen, Z. Liu, X.-Y. Huang, and X. Pan, 2022: The regularized WSM6 microphysical scheme and its validation in WRF 4D-Var. Adv. Atmos. Sci., in press.
TYPE: Enhancement KEYWORDS: WRFDA, AHI, Himawari-8 SOURCE: Craig Schwartz (NCAR/MMM) DESCRIPTION OF CHANGES: This PR makes several enhancements for assimilating Himawari-8 radiance data, including (1) Introduction of an all-sky obs error model (Harnisch et al., 2016) for all-sky AHI DA; (2) Optional read and use of AHI level-2 product (e.g., cloud mask); (3) More efficient read of a sub-area of the full disk data; (4) Allow the use of offline statistics of constant bias correction values; (5) More diagnostic output (peak of weighting functions, cloud ice water path, cloud flag) in omb/oma files. LIST OF MODIFIED FILES: var/da/da_radiance/module_radiance.f90 Registry/registry.var var/da/da_define_structures/da_define_structures.f90 var/da/da_monitor/da_rad_diags.f90 var/da/da_radiance/da_allocate_rad_iv.inc var/da/da_radiance/da_deallocate_radiance.inc var/da/da_radiance/da_get_innov_vector_crtm.inc var/da/da_radiance/da_initialize_rad_iv.inc var/da/da_radiance/da_qc_ahi.inc var/da/da_radiance/da_radiance.f90 var/da/da_radiance/da_radiance1.f90 var/da/da_radiance/da_radiance_init.inc var/da/da_radiance/da_read_obs_netcdf4ahi_jaxa.inc var/da/da_radiance/da_write_iv_rad_ascii.inc var/da/da_radiance/module_radiance.f90 var/run/ahi_info var/run/radiance_info/himawari-8-ahi.info TESTING: (1) WRFDA regression test passed; (2) AHI all-sky DA runs also Ok. Release Note: Enhancements for AHI radiance DA, including all-sky observation error model, Level-2 AHI product read, and more diagnostic output. Xu, D. M., Z. Q. Liu, S. Y. Fan, M. Chen, and F. F. Shen, 2021: Assimilating all-sky infrared radiances from Himawari-8 using the 3DVar method for the prediction of a severe storm over North China. Adv. Atmos. Sci., 38(4), 661-676.
This update is to resolve different simulation results on aarch64 and x86 for the same dataset (caused by round up error in the last position for some transcendental functions) TYPE: bug fix KEYWORDS: round up error, transcendental, compiler optimization SOURCE: Jun Tang, Amazon DESCRIPTION OF CHANGES: Problem: After investigation, different simulation results on aarch64 and x86 for the same dataset with gcc 10.2 are caused by round up error in the last position for some transcendental functions; and several optimization flags (FMA, inverse square root, ...). Solution: Use double precision for the computation of transcendental function at selected locations can make sure exact same output models are produced on aarch64 and x86 (this patch only covers a few PBL and cumulus scheme). Also disabling some risky optimization can guarantee same output as lower optimization level. LIST OF MODIFIED FILES: list of changed files (use `git diff --name-status master` to get formatted list) M arch/configure.defaults M phys/module_cu_tiedtke.F M phys/module_sf_myjsfc.F TESTS CONDUCTED: 1. The mods fixes the correctness problem between aarch64 and x86 for two WRF models (conus2.5km and conus12km). The output models are matched, bit by bit, on the two specified platform with the patch. 2. The regression tests have passed - as indicated the change should not affect its results. RELEASE NOTE: Fix numerical divergence on x86 and arm64. For best performance for WRF on arm64 please use armclang. (https://github.com/juntangc/notes/blob/main/release-note.pdf).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The first line should be a single-line "purpose" for this change
TYPE: choose one of [bug fix, enhancement, new feature, feature removed, no impact, text only]
KEYWORDS: 5 to 10 words related to commit, separated by commas
SOURCE: Either "developer's name (affiliation)" .XOR. "internal" for a WRF Dev committee member
DESCRIPTION OF CHANGES:
Problem:
Generally or specifically, what was wrong and needed to be addressed?
Solution:
What was down algorithmically and in the source code to address the problem?
ISSUE: For use when this PR closes an issue.
Fixes #123
LIST OF MODIFIED FILES: list of changed files (use
git diff --name-status masterto get formatted list)TESTS CONDUCTED:
RELEASE NOTE: Include a stand-alone message suitable for the inclusion in the minor and annual releases. A publication citation is appropriate.