Pull request for compiling shared objects.#47
Conversation
Merging Hotfixes branch HF_ounfpart into the WW3 master. * Bugfix in ww3_ounf.ftn: NBIPART check against NOSWLL was not accouting for zeroth wind partition. See Issue #2. * Updating link.wcoss_phase2 for grib compile, removing !$ lines in ww3_uprstr that broke compilation of codes with OMPH OMPG switch flags, added new global wave model switch. Added option for writing to fixed netcdf gridded output file when using NCO switch (the issue will be fixed more adequately in v7 following Issue NOAA-EMC#8). * DB1 depth-breaking_bugfix: added proper scaling coefficients for radian freq. [radHz]. See issue NOAA-EMC#7. Merge will lead to v6.07.1, tag and release will be updated accordingly.
Feature/cmakebased
Stelios' develop to master
…p ${CMAKE_SOURCE_DIR}/WW3/model -c theia_so -s NCEP_st4" line was updated
model/bin/cmplr.env : "optc='-c -module $path_m -no-fma -fPIC -ip -p -g -i4 -real-size 32 -fp-model precise -assume byterecl -convert big_endian -fno-alias -fno-fnalias'" was updated, "-fPIC"
make_makefile.sh: libww3.so was added
w3_automake: libww3.so was added
w3_make: libww3.so was added
mickaelaccensi
left a comment
mickaelaccensi
left a comment
There was a problem hiding this comment.
Hi all,
I would rather prefer to add it as a prefix so_intel to keep it consistent with options added in specific case or for specfic machine as it was done for zeus and datarmor.
It should be also done for the mpt and pgi compilers.
Mickael
JessicaMeixner-NOAA
left a comment
JessicaMeixner-NOAA
left a comment
There was a problem hiding this comment.
two minor typos.
| fi | ||
|
|
||
| echo -e "\n\n Files were copied from the data_regtests to the regtets folder." | ||
| echo -e "\n\n Files were copied from the data_regtests to the regstets folder." |
There was a problem hiding this comment.
@JessicaMeixner-NOAA at least I counted the number of letters 😄
|
@flampouris thanks for the pull request. Could you address comments by @mickaelaccensi: Would you be able to change intel_so to so_intel, and to provide options for mpt and pgi compilers? Thanks! |
|
The updates for intel and gnu will happen when we have a hpc available. I
won't be able to make updates that I am not familiar with.
…On Tue, Jun 11, 2019 at 10:45 AM Jose-Henrique Alves < ***@***.***> wrote:
@flampouris <https://github.com/flampouris> thanks for the pull request.
Could you address comments by @mickaelaccensi
<https://github.com/mickaelaccensi>: Would you be able to change intel_so
to so_intel, and to provide options for mpt and pgi compilers? Thanks!
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#47?email_source=notifications&email_token=AB7DH3ZV2T7ENDVLXWMWJHTPZ627BA5CNFSM4HW2M352YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXNL3IY#issuecomment-500874659>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AB7DH3ZLKJQDH7H7IBF4R7TPZ627BANCNFSM4HW2M35Q>
.
--
Dr. Rer. Nat. Stylianos Flampouris
Team Leader for Marine Data Assimilation
IMSG at NOAA/NWS/NCEP/EMC
*P: *+1(301)683-3789 *E: *stylianos.flampouris@noaa.gov
|
1. The _so extra suffix became an extension of the prefix 2. Typos at cmpl.env and ww3_ftp[...].sh
|
The code is updated according to the reviewers' suggestion. The updates were tested with Intel 14.* and 17.1 on Theia and the NCEP_st4 switch. @ajhenrique and @aliabdolali please let me know if there are any issues whiling testing. |
|
@flampouris please note that your pull request has been accepted into the branch FB_libso NOAA-EMC/WW3 for further testing before it is actually merged into the develop branch. We will let you know when that happens. Thanks! |
|
Thank you for the update.
<stylianos.flampouris@noaa.gov>
|
For completion of information justifying the rationale for asking this new pull request, please see comment #29 (comment), which provides results indicating a substantial increase in compile time when using the -fpic shared-object flag under the Intel compiler. This seems to be machine and/or compiler version dependent (so should be checked by developers that wish to run a large number of cases such as using a regtests matrix). |
…erturbations (NOAA-EMC#386) * Update .gitmodules and submodule pointer for fv3atm for gsl/develop branch * RUC ice for gsl/develop (replaces NOAA-EMC#47) (NOAA-EMC#49)Implementation of RUC LSM ice model in CCPP * Squash-merge climbfuji:rucice_gfsv16dzmin into gsl/develop * Add kice=9 to tests/tests/fv3_ccpp_rap and tests/tests/fv3_ccpp_hrrr * Change NEW_BASELINE directory for gsl/develop to avoid conflicts with development work on the authoritative branches * Add KICE=9 to tests/tests/fv3_ccpp_gsd_unified_ugwp and tests/tests/fv3_ccpp_gsd_drag_suite_unified_ugwp * Revert change to .gitmodules and update submodule pointer for fv3atm * Update gsl/develop from develop 2020/12/08 (NOAA-EMC#50) * Updates to stochastic_physics_wrapper (NOAA-EMC#280) Fix to stochastic_physics_wrapper to allow for random patterns to update at a longer time-step than model * Update for Jet, bug fixes in running with frac_grid=T and GFDL MP, and in restarting with frac_grid=T (NOAA-EMC#304) Update the modulefile for jet.intel to enable UPP v10.0.0. The hpc-stack v1.0.0 pre-release is used for this. Small changes are made to tests.rt.sh for jet.intel and gaea.intel (consistency with other platforms). The submodule pointer update for fv3atm addresses bugs in the ufs-weather-model with frac_grid=T and GFDL microphysics, and with restarting the model when frac_grid=T (from @shansun6 and @SMoorthi-emc). * Land stochastic perturbations (NOAA-EMC#57)
4 participants
This pull request is a response to #46 and continuation of #29. The compilation of shared objects becomes optional.
How to use:
./model/bin/w3_setup model -c intel_so -s NCEP_st4
cd [bin]
./w3_make libww3.so
See the code for details.
Several comparative tests showed that the average difference in compilation time for the master, with and without the options for shared objects is less than 0.2s. The original pull request had approximately the same delay (< 0.4s) in comparison with the master's compilation time.
The code has been tested with intel 17.1.