4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
5 |
============================== |
============================== |
6 |
|
|
7 |
|
o coupling interface pkgs: |
8 |
|
[atm_compon_interf, ocn_compon_interf & atm_ocn_coupler] |
9 |
|
- change the config check to test the agreement of Ocean Depth |
10 |
|
with ATM land/sea mask (instead of ATM orography); |
11 |
|
- send to coupler the number of coupler time-steps to do for this run: |
12 |
|
this allows to check for consistency and don't need anymore to be |
13 |
|
specified in parameter file "data"; |
14 |
|
in coupler: |
15 |
|
- check consistency of RunOff map versus Ocean Depth and ATM land/sea mask. |
16 |
|
- improve report to log-file of config setting (pass log-file unit to few S/R) |
17 |
|
o pkg/compon_communic: |
18 |
|
- add coupling send/receive for non-tiled integer vector; |
19 |
|
- add S/R to check and stop if any proc has an error (similar to |
20 |
|
eesupp/src/stop_if_error.F but single thread and across Comm_World). |
21 |
|
o pkg/exch2: |
22 |
|
- in exch2*_get_rx[1,2]: use new list "W2_tileIndex" to replace search |
23 |
|
through list of all tiles my proc owns. |
24 |
|
o model/src: |
25 |
|
- AUTODIFF only: in calc_adv_flow.F, recompute rTransKp (instead of copying |
26 |
|
from rTrans) to simplify dependencies (no need to store rTrans anymore in |
27 |
|
calling S/R); this affects AD results at machine truncation level: |
28 |
|
update AD-output (TAF) of experiments: 1D_ocean_ice_column, |
29 |
|
bottom_ctrl_5x5 & obcs_ctrl (12 digits for AD grad, same FWD grad). |
30 |
|
o tools/build_options: |
31 |
|
- simplify Warnings options (-ieee & -devel cases) in linux_amd64_gfortran |
32 |
|
o coupling interface pkgs: |
33 |
|
[compon_communic, atm_compon_interf & ocn_compon_interf] |
34 |
|
- move declaration of local array "ibuf" out of header file "CPLR_SIG.h" |
35 |
|
to each S/R that needs it; remove un-used variables. |
36 |
|
- change component tile registration (cpl_register.F in both compon_interf |
37 |
|
and mitcomponent_tile_register.F) to pass the full list of tiles |
38 |
|
information (including bi,bj) in 1 vector (instead of in a bi,bj matrix). |
39 |
|
- with pkg/exch2: adopt the same convention used for I/O to define mapping |
40 |
|
towards coupler global array (unless W2_useE2ioLayOut=F): |
41 |
|
a) can now use different tiling between atm & ocn comp. |
42 |
|
b) no needs to modify the run-Off map when changing tiling (was needed |
43 |
|
even if using the same tiling between atm & ocn). |
44 |
|
c) will allow to run with blank-tiles in 1 component (e.g., for ocean |
45 |
|
full land tiles); not yet tested. |
46 |
|
- check for valid mapping from comp. tile to coupler global array |
47 |
|
- improve report of tile registration results to log file. |
48 |
|
o pkg/exch2: |
49 |
|
- in eesupp/src/eeboot.F: move call to W2_EEBOOT before Coupler |
50 |
|
tile-registration (CPL_REGISTER); |
51 |
|
- activate (un-comment) definition and setting of "W2_tileIndex" list to |
52 |
|
store the local bi,bj indices of each tile; |
53 |
|
- also print bi,bj in local report to log file; |
54 |
|
- fix pkg/debug/debug_cs_corner_uv.F for the case nSy > 1 (left from |
55 |
|
Jun 2009 modif of pkg/exch2). |
56 |
|
o testing on baudelaire: |
57 |
|
- run 2 tests at a time: no significant slow down (since enough cpu) and |
58 |
|
get the full set of test done earlier |
59 |
|
- switch the ifort test to "-devel" ; run the OpenAD test everyday. |
60 |
|
o model/src/thermodynamics.F and routines called from it: |
61 |
|
- pkg/gmredi: gmredi_residual_flow.F (replacing gmredi_calc_uvflow.F & |
62 |
|
gmredi_calc_wflow.F) to update eulerian velocity (input) into residual |
63 |
|
velocity (outp) by adding bolus velocity. |
64 |
|
- model/src: calc_adv_flow.F (replacing calc_common_factors.F) to compute |
65 |
|
mass transport from velocity + lateral grid cell area |
66 |
|
- rename calc_gt.F to temp_integrate.F (includes call to TIMESTEP_TRACER) |
67 |
|
- rename calc_gs.F to salt_integrate.F (includes call to TIMESTEP_TRACER) |
68 |
|
in thermodynamics.F |
69 |
|
- remove k from thermodynamics.F and move it to temp_integrate.F |
70 |
|
salt_integrate.F and ptracers_integrate.F (now done inside the tracer |
71 |
|
loop). |
72 |
|
- compute locally (in thermodynamics.F) 3-D velocity field that is used to |
73 |
|
advect tracers; pass it as argument to GAD_ADVECTION, GAD_SOM_ADVECT, |
74 |
|
PTRACERS_ADVECTION, TEMP_INTEGRATE, SALT_INTEGRATE, PTRACERS_INTEGRATE, |
75 |
|
GAD_IMPLICIT_R and PTRACERS_IMPLICIT |
76 |
|
- update AD-output (TAF & OAD) of experiments: 1D_ocean_ice_column, |
77 |
|
global_ocean.90x40x15, tutorial_global_oce_biogeo and |
78 |
|
tutorial_tracer_adjsens (+ TLM output). |
79 |
|
o tools/OAD_support and (verification/*/) code_oad: |
80 |
|
- OAD_support/cb2mGetModules.csh: add PACKAGES_CONFIG.h in header file to |
81 |
|
module conversion ; remove unneeded CPP_EEOPTIONS.h. |
82 |
|
- code_oad/CPP_OPTIONS.h: remove PACKAGES_CONFIG.h and use a more standard |
83 |
|
version (closer to model/inc/CPP_OPTIONS.h). |
84 |
|
o model/src: |
85 |
|
- remove parameter "useOldFreezing" from PARAMS.h and push it to the |
86 |
|
retired list of parameters (routine FREEZE has been replace by surface |
87 |
|
freezing (10.yrs ago) and by pkg/frazil (interior freezing). |
88 |
|
o pkg/gmredi: |
89 |
|
- add option (GM_CALC_TRANSP_FROM_VEL) to compute transport directly from |
90 |
|
velocity (same scheme but affects machine truncation) for GM_AdvForm. |
91 |
|
- update output of FWD experiments: front_relax.bvp, |
92 |
|
global_ocean.90x40x15.dwnslp, global_ocean.cs32x15 and ideal_2D_oce. |
93 |
|
|
94 |
|
checkpoint64q (2013/11/17) |
95 |
|
o verification: |
96 |
|
- generate new output on baudelaire (gfortran, -ieee) for OpenAD experiments |
97 |
|
tutorial_global_oce_biogeo and global_ocean.90x40x15. |
98 |
|
- generate new output (growth & melt) for seaice_itd experiment after |
99 |
|
updating the input pickup_seaice files (previously wrong). |
100 |
|
o pkg/cost: |
101 |
|
- define a "final cost function" for each tile, and use GLOBAL_SUM_TILE_RL |
102 |
|
(instead of _GLOBAL_SUM_RL) to get the final cost from all procs & threads: |
103 |
|
o this fixes global-sum value if multi-threads (since GLOBAL_SUM_TILE_RL |
104 |
|
cannot be applied to var in common block). |
105 |
|
o allows to get results independent of tile-per-proc setting (if using |
106 |
|
slower GLOBAL_SUM_SEND_RECV option). |
107 |
|
- update accordingly: seaice_cost_final.F, shelfice_cost_final.F, |
108 |
|
thsice_cost_final.F and ecco_cost_final.F (this one affects lab_sea |
109 |
|
forward gradients: update output). |
110 |
|
- update */code_oad/externalDummies.F by adding dummy GLOBAL_SUM_TILE_RL |
111 |
|
S/R and, for most exp., also comment out GLOBAL_SUM_R8. |
112 |
|
o pkg/mom_common: |
113 |
|
- move to pkg/mom_common and model/src (previously in tutorial_deep_convection |
114 |
|
code) 2nd version of isotropic 3-D Smagorinsky code interface: strain and |
115 |
|
viscosity are locally declared in dynmics.F and pass as argument to CALC_GW; |
116 |
|
ensure that all field value that are used are set. |
117 |
|
o tutorial_deep_convection (smag3d): |
118 |
|
- add missing scaling in mom_calc_smag_3d.F; |
119 |
|
- set smag3D_coeff in input.smag3d such that results do not change. |
120 |
|
o pkg/dic: |
121 |
|
- change dic_cost.F to compute cost on each tile first and then apply |
122 |
|
GLOBAL_SUM_TILE_RL: this affects forward gradient of AD experiments |
123 |
|
tutorial_dic_adjoffline & tutorial_global_oce_biogeo; update output. |
124 |
|
- remove kLev argument and use GLOBAL_SUM_TILE_RL in tracer_meanarea. |
125 |
|
o eesupp/src & pkg/mdsio: |
126 |
|
- remove gather/scatter_vector.F (from eesupp) and mdsio_read/writevector.F |
127 |
|
(from pkg/mdsio) since they are no longer used for tape IO. |
128 |
|
- move simple C subroutine "setdir.c" from pkg/compon_communic to eesupp/src |
129 |
|
(+ keep a copy in pkg/atm_ocn_coupler). |
130 |
|
o verification/isomip: |
131 |
|
- split domain in 8 squared tiles (instead of just 2) for FWD, AD & OAD tests. |
132 |
|
o tools/build_options: |
133 |
|
- optfile linux_amd64_ifort11, fast (no ieee) case: |
134 |
|
to get a working restart for all experiments, add "-fp-model source" |
135 |
|
to FOPTIM (also remove "-opt-streaming-stores auto" <- this is the default). |
136 |
|
|
137 |
|
checkpoint64p (2013/10/20) |
138 |
|
o pkg/autodiff: |
139 |
|
- adread_adwrite.F: when type and length do not match ALLOW_AUTODIFF_WHTAPEIO |
140 |
|
criteria, return to simple (default) tape IO format. |
141 |
|
o pkg/mdsio (+ eesupp/src): |
142 |
|
- rewrite/simplify tape-IO S/R (new S/R MDS_READ/WRITE_TAPE replace |
143 |
|
previous MDSREAD/WRITEVECTOR) with 2 array argument from each type (R4/R8); |
144 |
|
- fix globalFile and singleCpuIO options using simpler global mapping. |
145 |
|
- new pairs of scatter/gather vector S/R (one for each type, generated from |
146 |
|
template) corresponding to simpler global mapping. |
147 |
|
- adread_write.F: switch to new S/R (but leave old calls commented) and write |
148 |
|
to globalFile when parameter globalFiles is set to True. |
149 |
|
- switch on useSingleCpuIO in AD experiments: global_ocean.90x40x15 & isomip. |
150 |
|
o pkg/autodiff: |
151 |
|
- adread_adwrite.F: add a hack for RS arrays (when RS=real*4); |
152 |
|
with ALLOW_AUTODIFF_WHTAPEIO defined, since corresponding (real*4) is |
153 |
|
missing, just add a STOP if array type do not match. |
154 |
|
o pkg/exf: |
155 |
|
- to allow to use seaice-fraction from pkg/exf in other pkgs than pkg/seaice: |
156 |
|
rename CPP option EXF_ALLOW_SEAICE_RELAX to EXF_SEAICE_FRACTION |
157 |
|
and rename array obsSIce to exf_iceFraction. |
158 |
|
o tutorial_deep_convection: |
159 |
|
- add 1rst version of isotropic 3-D Smagorinsky code (from Louis-Philippe), |
160 |
|
for now all in this exp. code dir. |
161 |
|
- add secondary test using Smag-3D viscosity (keep very low backgroud visc for |
162 |
|
testing) and flux-limter advection scheme for Temp. (no explicit diffusion). |
163 |
|
o eesupp/src: |
164 |
|
- added SINGLE_DISK_IO option because reading of parameter files on |
165 |
|
very large number of processors did not work with existing set-up |
166 |
|
|
167 |
|
checkpoint64o (2013/09/25) |
168 |
|
o pkg/atm_phys & atm_compon_interf: |
169 |
|
- allows for coupling of gray-atmosphere (atm_phys) with ocean. |
170 |
|
o tools/build_options: |
171 |
|
- switch from "-fPIC" to "-mcmodel=medium" in optfiles: linux_amd64_gfortran |
172 |
|
& linux_amd64_ifort11 and in linux_amd64_g77 & linux_amd64_open64 also |
173 |
|
remove CPP setting (was identical to genmake2 default). |
174 |
|
o pkg/gmredi & model/src: Some updating and work done on the residual model. |
175 |
|
- created new file model/src/calc_eddy_stress.F to calculate the eddy stress. |
176 |
|
- fixed up the calculation of the eddy stress and how it is added to the |
177 |
|
external forcing via taueddy_external_forcing.F |
178 |
|
- at the moment, it is very experimental and only works with the variable eddy |
179 |
|
diffusivity calculated in gmredi_k3d.F |
180 |
|
o pkg/seaice/seaice_check.F: add a check to catch the case of specifying |
181 |
|
lw/swfluxfile instead of lw/swdownfile in data.exf together with pkg/seaice |
182 |
|
o testreport (& tools/do_tst_2+2): |
183 |
|
- for MPI test, set default COMMAND to 'mpirun -v -np TR_NPROC ./{executable}' |
184 |
|
o tools/build_options: linux_amd64_gfortran & linux_amd64_ifort11: |
185 |
|
- remove CPP='cpp -traditional -P' (this is genmake2 default) |
186 |
|
- add switch (for big setups) to use 'medium' memory-model (but leave it |
187 |
|
commented out for now). Note: needed with FC 19 to compile big setups. |
188 |
|
o pkg/atm_compon_interf: |
189 |
|
- regroup per field origin (model & pkgs) the coupling storage routines |
190 |
|
(which accumalate in time each coupling field) |
191 |
|
- move bi,bj loops inside atm_store_my_data.F and store also wind-stress; |
192 |
|
update accordingly + simplify aim_do_physics.F |
193 |
|
|
194 |
|
checkpoint64n (2013/08/26) |
195 |
|
o pkg/bbl: |
196 |
|
- update formulation to reduce dispersion of tracer properties in bbl |
197 |
|
- replaced horizontal and vertical transport parameters with velocities |
198 |
|
- changes results for verification/global_with_exf.yearly |
199 |
|
o verification/halfpipe_streamice: |
200 |
|
- change to matrix construction; update both Fwd & AD output. |
201 |
|
o verification/tutorial_tracer_adjsens: |
202 |
|
- OpenAD test: switch to vector-invariant momentum since fluxform with |
203 |
|
rStar causes problems with OpenAD; update output. |
204 |
|
o tools: |
205 |
|
- genmake2: define a common prefix for all temp-file names ; move them |
206 |
|
to local /tmp disk (but would be easy to move them back); |
207 |
|
- xmakedepend: apply the same switch for temp-files. |
208 |
|
o pkg/offline: |
209 |
|
- changes for use with rStar (NLFS): assume that we continue to provide |
210 |
|
UVELMASS/hFacW_ini & VVELMASS/hFacS_ini (hFacWi,S_ini as written in grid |
211 |
|
output files) and apply r* scaling at each time-step (function of the time |
212 |
|
varying Eta field); use normal pickup files (needed for restart with NLFS). |
213 |
|
o adjoint with OpenAD: |
214 |
|
- move OpenAD-specific files from verification/OpenAD/code_oad_all to a new |
215 |
|
package openad. This package has to be compiled for all runs with OpenAD |
216 |
|
instead of specifying code_oad_all as a modification directory. |
217 |
|
- link the fortran templates from tools/OAD_support to the build directory |
218 |
|
for preprocessing to avoid conflicts between different setups compiled |
219 |
|
simultaneously and simplify cleanup. |
220 |
|
o pkg/exf: |
221 |
|
- in exf_getforcing.F: comment out most of the EXCH calls (fields used by |
222 |
|
main model or other pkgs are exchanged in EXF_MAPFIELDS). |
223 |
|
o model/src: |
224 |
|
- do_stagger_fields_exchanges.F: skip EXCH calls when using pkg/offline |
225 |
|
(similar to do_fields_blocking_exchanges.F). |
226 |
|
o tools/genmake2: |
227 |
|
- remove TAF output src code just before calling staf (when not doing a fresh |
228 |
|
built, will stop if things go wrong with staf). |
229 |
|
- fix target "adobj" (left from TAF version-2 transition) |
230 |
|
o pkg/diagnostics: |
231 |
|
- add a parameter to track the status of pkg/diagnostics activation; |
232 |
|
this replace/extend the use of logical param "settingDiags" (to check |
233 |
|
when adding diag to the list) to also check any DIAGNOSTICS_[]FILL* call. |
234 |
|
New S/R to print error message and stop when pkgStatus is not right. |
235 |
|
- to try: add a warning when "data.diagnostics" exist but useDiagnostics=F. |
236 |
|
|
237 |
|
checkpoint64m (2013/08/13) |
238 |
|
o model/inc: |
239 |
|
- move FCORI_MACROS.h from PARAMS(_MACROS).h to GRID_MACROS.h (left from |
240 |
|
8 yrs ago modif when Coriolis arrays were moved from PARAMS.h to GRID.h) |
241 |
|
o pkg/kpp & aim_v23: |
242 |
|
- only call DIAGNOSTICS_FILL when useDiagnostics=TRUE. |
243 |
|
o pkg/seaice: |
244 |
|
- restore (last time on 2011/12/24) ability to compile pkg/seaice with |
245 |
|
undefined pkg/exf ALLOW_ATM_TEMP or ALLOW_RUNOFF CPP-options. |
246 |
|
- only call DIAGNOSTICS_FILL when useDiagnostics=TRUE. |
247 |
|
o eesupp/src: |
248 |
|
- fix check for overlap-size argument (myOLn.GT.MAX_OLY_EXCH) |
249 |
|
in exch1_rx[_ad].template (thanks to Francois Lefeuvre report). |
250 |
|
o switch to TAF version-2: |
251 |
|
- add flow directives to "freeze" naming convention for ad-routines that |
252 |
|
are used in I/O routines of adjoint state and associated common blocks |
253 |
|
that are used in adcommon.h (backward compatible with version-1) |
254 |
|
- update all ad-optfiles (in tools/adjoint_options + few local to some |
255 |
|
verification exp) to switch to latest TAF version (currently 2.5.5); |
256 |
|
merge "*_mpi" adoptfiles into the corresponding non-mpi (but commented |
257 |
|
out except for diva). |
258 |
|
- the major change with TAF version-2 is that both the differentiated |
259 |
|
routines and the forward routines (new) are returned as part of the TAF |
260 |
|
output source code. Update genmake2 accordingly. |
261 |
|
- new output for AD experiment tutorial_global_oce_biogeo (better AD gradient |
262 |
|
with version-2 when compared to TLM or FWD gradient); other experiments |
263 |
|
output show minor changes (machine truncation level); update the few AD & |
264 |
|
TLM output with 12 matching digits only (some prior to version-2 switch). |
265 |
|
|
266 |
|
checkpoint64l (2013/08/05) |
267 |
|
o model/src: |
268 |
|
- in dynamics.F, skip the call to CALC_VISCOSITY if momViscosity=F |
269 |
|
- main_do_loop.F: no longer includes additional (AUTODIFF) header files |
270 |
|
when using TAF but only if using OpenAD. |
271 |
|
o pkg/mom_common+fluxform+vecinv: |
272 |
|
- always set horiz. viscosity arrays to background value before calling |
273 |
|
MOM_CALC_VISC (in MOM_FLUXFORM & MOM_VECINV) and call S/R MOM_CALC_VISC |
274 |
|
only when using variable horiz. viscosity (useVariableVisc=T); |
275 |
|
- simplify mom_vecinv.F (only 1 block for momViscosity). |
276 |
|
o pkg/mom_common: variable horiz. viscosity in NH code |
277 |
|
- move horiz-viscosity 3-D arrays viscAh_W & viscA4_W (used in calc_gw.F) |
278 |
|
from model/inc/NH_VARS.h to pkg/mom_common/MOM_VISC.h |
279 |
|
- store in common block (in MOM_VISC.h) logical parameters: useHarmonicVisc, |
280 |
|
useBiharmonicVisc & useVariableVisc. |
281 |
|
- fix old bug (already in 1rst implementation of variable horizontal viscosity |
282 |
|
in calc_gw.F, version 1.27, 2006/06/20): |
283 |
|
horizontal bi-harmonic viscosity in gW was missing when wiscA4W=0 |
284 |
|
even if variable bi-harmonic viscosity (e.g., viscA4Grid) was non-zero. |
285 |
|
- change background horiz viscosity in vertical momentum (gW) to always be |
286 |
|
viscAhW,viscA4W (instead of switching to viscAhD, viscA4D when using |
287 |
|
any type of variable viscosity, e.g., viscAh/4Grid) |
288 |
|
- change default value of viscAhW,viscA4W to viscAhD,viscA4D (instead of |
289 |
|
viscAh,viscA4) which default to viscAh,viscA4 : this avoid missing |
290 |
|
horiz. viscosity in W eq when just setting viscAhD/Z,viscA4D/Z (since |
291 |
|
default viscAh,viscA4 is zero). |
292 |
|
o model/src: |
293 |
|
- extend length of drC & recip_drC from Nr to Nr+1. |
294 |
o Divided Adjoint (DIVA) with MPI: |
o Divided Adjoint (DIVA) with MPI: |
295 |
- genmake2: do not create local ./mpi_headers dir when no file to copy into. |
- genmake2: do not create local ./mpi_headers dir when no file to copy into. |
296 |
- remove option ALLOW_DIVIDED_ADJOINT_MPI (replaced by ALLOW_DIVIDED_ADJOINT |
- remove option ALLOW_DIVIDED_ADJOINT_MPI (replaced by ALLOW_DIVIDED_ADJOINT |