| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o pkg/exch2: |
| 8 |
|
- in exch2*_get_rx[1,2]: use new list "W2_tileIndex" to replace search |
| 9 |
|
through list of all tiles my proc owns. |
| 10 |
|
o model/src: |
| 11 |
|
- AUTODIFF only: in calc_adv_flow.F, recompute rTransKp (instead of copying |
| 12 |
|
from rTrans) to simplify dependencies (no need to store rTrans anymore in |
| 13 |
|
calling S/R); this affects AD results at machine truncation level: |
| 14 |
|
update AD-output (TAF) of experiments: 1D_ocean_ice_column, |
| 15 |
|
bottom_ctrl_5x5 & obcs_ctrl (12 digits for AD grad, same FWD grad). |
| 16 |
|
o tools/build_options: |
| 17 |
|
- simplify Warnings options (-ieee & -devel cases) in linux_amd64_gfortran |
| 18 |
|
o coupling interface pkgs: |
| 19 |
|
[compon_communic, atm_compon_interf & ocn_compon_interf] |
| 20 |
|
- move declaration of local array "ibuf" out of header file "CPLR_SIG.h" |
| 21 |
|
to each S/R that needs it; remove un-used variables. |
| 22 |
|
- change component tile registration (cpl_register.F in both compon_interf |
| 23 |
|
and mitcomponent_tile_register.F) to pass the full list of tiles |
| 24 |
|
information (including bi,bj) in 1 vector (instead of in a bi,bj matrix). |
| 25 |
|
- with pkg/exch2: adopt the same convention used for I/O to define mapping |
| 26 |
|
towards coupler global array (unless W2_useE2ioLayOut=F): |
| 27 |
|
a) can now use different tiling between atm & ocn comp. |
| 28 |
|
b) no needs to modify the run-Off map when changing tiling (was needed |
| 29 |
|
even if using the same tiling between atm & ocn). |
| 30 |
|
c) will allow to run with blank-tiles in 1 component (e.g., for ocean |
| 31 |
|
full land tiles); not yet tested. |
| 32 |
|
- check for valid mapping from comp. tile to coupler global array |
| 33 |
|
- improve report of tile registration results to log file. |
| 34 |
|
o pkg/exch2: |
| 35 |
|
- in eesupp/src/eeboot.F: move call to W2_EEBOOT before Coupler |
| 36 |
|
tile-registration (CPL_REGISTER); |
| 37 |
|
- activate (un-comment) definition and setting of "W2_tileIndex" list to |
| 38 |
|
store the local bi,bj indices of each tile; |
| 39 |
|
- also print bi,bj in local report to log file; |
| 40 |
|
- fix pkg/debug/debug_cs_corner_uv.F for the case nSy > 1 (left from |
| 41 |
|
Jun 2009 modif of pkg/exch2). |
| 42 |
|
o testing on baudelaire: |
| 43 |
|
- run 2 tests at a time: no significant slow down (since enough cpu) and |
| 44 |
|
get the full set of test done earlier |
| 45 |
|
- switch the ifort test to "-devel" ; run the OpenAD test everyday. |
| 46 |
|
o model/src/thermodynamics.F and routines called from it: |
| 47 |
|
- pkg/gmredi: gmredi_residual_flow.F (replacing gmredi_calc_uvflow.F & |
| 48 |
|
gmredi_calc_wflow.F) to update eulerian velocity (input) into residual |
| 49 |
|
velocity (outp) by adding bolus velocity. |
| 50 |
|
- model/src: calc_adv_flow.F (replacing calc_common_factors.F) to compute |
| 51 |
|
mass transport from velocity + lateral grid cell area |
| 52 |
|
- rename calc_gt.F to temp_integrate.F (includes call to TIMESTEP_TRACER) |
| 53 |
|
- rename calc_gs.F to salt_integrate.F (includes call to TIMESTEP_TRACER) |
| 54 |
|
in thermodynamics.F |
| 55 |
|
- remove k from thermodynamics.F and move it to temp_integrate.F |
| 56 |
|
salt_integrate.F and ptracers_integrate.F (now done inside the tracer |
| 57 |
|
loop). |
| 58 |
|
- compute locally (in thermodynamics.F) 3-D velocity field that is used to |
| 59 |
|
advect tracers; pass it as argument to GAD_ADVECTION, GAD_SOM_ADVECT, |
| 60 |
|
PTRACERS_ADVECTION, TEMP_INTEGRATE, SALT_INTEGRATE, PTRACERS_INTEGRATE, |
| 61 |
|
GAD_IMPLICIT_R and PTRACERS_IMPLICIT |
| 62 |
|
- update AD-output (TAF & OAD) of experiments: 1D_ocean_ice_column, |
| 63 |
|
global_ocean.90x40x15, tutorial_global_oce_biogeo and |
| 64 |
|
tutorial_tracer_adjsens (+ TLM output). |
| 65 |
|
o tools/OAD_support and (verification/*/) code_oad: |
| 66 |
|
- OAD_support/cb2mGetModules.csh: add PACKAGES_CONFIG.h in header file to |
| 67 |
|
module conversion ; remove unneeded CPP_EEOPTIONS.h. |
| 68 |
|
- code_oad/CPP_OPTIONS.h: remove PACKAGES_CONFIG.h and use a more standard |
| 69 |
|
version (closer to model/inc/CPP_OPTIONS.h). |
| 70 |
|
o model/src: |
| 71 |
|
- remove parameter "useOldFreezing" from PARAMS.h and push it to the |
| 72 |
|
retired list of parameters (routine FREEZE has been replace by surface |
| 73 |
|
freezing (10.yrs ago) and by pkg/frazil (interior freezing). |
| 74 |
|
o pkg/gmredi: |
| 75 |
|
- add option (GM_CALC_TRANSP_FROM_VEL) to compute transport directly from |
| 76 |
|
velocity (same scheme but affects machine truncation) for GM_AdvForm. |
| 77 |
|
- update output of FWD experiments: front_relax.bvp, |
| 78 |
|
global_ocean.90x40x15.dwnslp, global_ocean.cs32x15 and ideal_2D_oce. |
| 79 |
|
|
| 80 |
|
checkpoint64q (2013/11/17) |
| 81 |
|
o verification: |
| 82 |
|
- generate new output on baudelaire (gfortran, -ieee) for OpenAD experiments |
| 83 |
|
tutorial_global_oce_biogeo and global_ocean.90x40x15. |
| 84 |
|
- generate new output (growth & melt) for seaice_itd experiment after |
| 85 |
|
updating the input pickup_seaice files (previously wrong). |
| 86 |
|
o pkg/cost: |
| 87 |
|
- define a "final cost function" for each tile, and use GLOBAL_SUM_TILE_RL |
| 88 |
|
(instead of _GLOBAL_SUM_RL) to get the final cost from all procs & threads: |
| 89 |
|
o this fixes global-sum value if multi-threads (since GLOBAL_SUM_TILE_RL |
| 90 |
|
cannot be applied to var in common block). |
| 91 |
|
o allows to get results independent of tile-per-proc setting (if using |
| 92 |
|
slower GLOBAL_SUM_SEND_RECV option). |
| 93 |
|
- update accordingly: seaice_cost_final.F, shelfice_cost_final.F, |
| 94 |
|
thsice_cost_final.F and ecco_cost_final.F (this one affects lab_sea |
| 95 |
|
forward gradients: update output). |
| 96 |
|
- update */code_oad/externalDummies.F by adding dummy GLOBAL_SUM_TILE_RL |
| 97 |
|
S/R and, for most exp., also comment out GLOBAL_SUM_R8. |
| 98 |
|
o pkg/mom_common: |
| 99 |
|
- move to pkg/mom_common and model/src (previously in tutorial_deep_convection |
| 100 |
|
code) 2nd version of isotropic 3-D Smagorinsky code interface: strain and |
| 101 |
|
viscosity are locally declared in dynmics.F and pass as argument to CALC_GW; |
| 102 |
|
ensure that all field value that are used are set. |
| 103 |
|
o tutorial_deep_convection (smag3d): |
| 104 |
|
- add missing scaling in mom_calc_smag_3d.F; |
| 105 |
|
- set smag3D_coeff in input.smag3d such that results do not change. |
| 106 |
|
o pkg/dic: |
| 107 |
|
- change dic_cost.F to compute cost on each tile first and then apply |
| 108 |
|
GLOBAL_SUM_TILE_RL: this affects forward gradient of AD experiments |
| 109 |
|
tutorial_dic_adjoffline & tutorial_global_oce_biogeo; update output. |
| 110 |
|
- remove kLev argument and use GLOBAL_SUM_TILE_RL in tracer_meanarea. |
| 111 |
|
o eesupp/src & pkg/mdsio: |
| 112 |
|
- remove gather/scatter_vector.F (from eesupp) and mdsio_read/writevector.F |
| 113 |
|
(from pkg/mdsio) since they are no longer used for tape IO. |
| 114 |
|
- move simple C subroutine "setdir.c" from pkg/compon_communic to eesupp/src |
| 115 |
|
(+ keep a copy in pkg/atm_ocn_coupler). |
| 116 |
|
o verification/isomip: |
| 117 |
|
- split domain in 8 squared tiles (instead of just 2) for FWD, AD & OAD tests. |
| 118 |
|
o tools/build_options: |
| 119 |
|
- optfile linux_amd64_ifort11, fast (no ieee) case: |
| 120 |
|
to get a working restart for all experiments, add "-fp-model source" |
| 121 |
|
to FOPTIM (also remove "-opt-streaming-stores auto" <- this is the default). |
| 122 |
|
|
| 123 |
|
checkpoint64p (2013/10/20) |
| 124 |
|
o pkg/autodiff: |
| 125 |
|
- adread_adwrite.F: when type and length do not match ALLOW_AUTODIFF_WHTAPEIO |
| 126 |
|
criteria, return to simple (default) tape IO format. |
| 127 |
|
o pkg/mdsio (+ eesupp/src): |
| 128 |
|
- rewrite/simplify tape-IO S/R (new S/R MDS_READ/WRITE_TAPE replace |
| 129 |
|
previous MDSREAD/WRITEVECTOR) with 2 array argument from each type (R4/R8); |
| 130 |
|
- fix globalFile and singleCpuIO options using simpler global mapping. |
| 131 |
|
- new pairs of scatter/gather vector S/R (one for each type, generated from |
| 132 |
|
template) corresponding to simpler global mapping. |
| 133 |
|
- adread_write.F: switch to new S/R (but leave old calls commented) and write |
| 134 |
|
to globalFile when parameter globalFiles is set to True. |
| 135 |
|
- switch on useSingleCpuIO in AD experiments: global_ocean.90x40x15 & isomip. |
| 136 |
|
o pkg/autodiff: |
| 137 |
|
- adread_adwrite.F: add a hack for RS arrays (when RS=real*4); |
| 138 |
|
with ALLOW_AUTODIFF_WHTAPEIO defined, since corresponding (real*4) is |
| 139 |
|
missing, just add a STOP if array type do not match. |
| 140 |
|
o pkg/exf: |
| 141 |
|
- to allow to use seaice-fraction from pkg/exf in other pkgs than pkg/seaice: |
| 142 |
|
rename CPP option EXF_ALLOW_SEAICE_RELAX to EXF_SEAICE_FRACTION |
| 143 |
|
and rename array obsSIce to exf_iceFraction. |
| 144 |
|
o tutorial_deep_convection: |
| 145 |
|
- add 1rst version of isotropic 3-D Smagorinsky code (from Louis-Philippe), |
| 146 |
|
for now all in this exp. code dir. |
| 147 |
|
- add secondary test using Smag-3D viscosity (keep very low backgroud visc for |
| 148 |
|
testing) and flux-limter advection scheme for Temp. (no explicit diffusion). |
| 149 |
|
o eesupp/src: |
| 150 |
|
- added SINGLE_DISK_IO option because reading of parameter files on |
| 151 |
|
very large number of processors did not work with existing set-up |
| 152 |
|
|
| 153 |
|
checkpoint64o (2013/09/25) |
| 154 |
|
o pkg/atm_phys & atm_compon_interf: |
| 155 |
|
- allows for coupling of gray-atmosphere (atm_phys) with ocean. |
| 156 |
|
o tools/build_options: |
| 157 |
|
- switch from "-fPIC" to "-mcmodel=medium" in optfiles: linux_amd64_gfortran |
| 158 |
|
& linux_amd64_ifort11 and in linux_amd64_g77 & linux_amd64_open64 also |
| 159 |
|
remove CPP setting (was identical to genmake2 default). |
| 160 |
|
o pkg/gmredi & model/src: Some updating and work done on the residual model. |
| 161 |
|
- created new file model/src/calc_eddy_stress.F to calculate the eddy stress. |
| 162 |
|
- fixed up the calculation of the eddy stress and how it is added to the |
| 163 |
|
external forcing via taueddy_external_forcing.F |
| 164 |
|
- at the moment, it is very experimental and only works with the variable eddy |
| 165 |
|
diffusivity calculated in gmredi_k3d.F |
| 166 |
|
o pkg/seaice/seaice_check.F: add a check to catch the case of specifying |
| 167 |
|
lw/swfluxfile instead of lw/swdownfile in data.exf together with pkg/seaice |
| 168 |
|
o testreport (& tools/do_tst_2+2): |
| 169 |
|
- for MPI test, set default COMMAND to 'mpirun -v -np TR_NPROC ./{executable}' |
| 170 |
|
o tools/build_options: linux_amd64_gfortran & linux_amd64_ifort11: |
| 171 |
|
- remove CPP='cpp -traditional -P' (this is genmake2 default) |
| 172 |
|
- add switch (for big setups) to use 'medium' memory-model (but leave it |
| 173 |
|
commented out for now). Note: needed with FC 19 to compile big setups. |
| 174 |
o pkg/atm_compon_interf: |
o pkg/atm_compon_interf: |
| 175 |
- regroup per field origin (model & pkgs) the coupling storage routines |
- regroup per field origin (model & pkgs) the coupling storage routines |
| 176 |
(which accumalate in time each coupling field) |
(which accumalate in time each coupling field) |
| 177 |
- move bi,bj loops inside atm_store_my_data.F and store also wind-stress; |
- move bi,bj loops inside atm_store_my_data.F and store also wind-stress; |
| 178 |
update accordingly + simplify aim_do_physics.F |
update accordingly + simplify aim_do_physics.F |
| 179 |
|
|
| 180 |
checkpoint64n (2013/08/26) |
checkpoint64n (2013/08/26) |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch