| 4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
| 5 |
============================== |
============================== |
| 6 |
|
|
| 7 |
|
o pkg/seaice: vectorize seaice_solve4temp.F |
| 8 |
|
- move iterations loops outside of i/j-loops |
| 9 |
|
- requires 2D arrays for TB and effConduct |
| 10 |
|
- add 2D iceOrNot to replace HICE_ACTUAL.GT.0, as HICE_ACTUAL is |
| 11 |
|
modified in this routine |
| 12 |
|
- a little bit of clean up, adjusting indentation, etc. |
| 13 |
|
- replaced a few TEN**a by exp(a*log(10.)) in new code for better |
| 14 |
|
performance and '**TWO' by '**2' (because TWO is a float), |
| 15 |
|
replaced ONE, ZERO |
| 16 |
|
o eesupp/src: |
| 17 |
|
fix flow dir for exch1_z_RX_cube, exch1_uv_RX_cube & exch1_bg_RX_cube |
| 18 |
|
o pkg/seaice/seaice_growth.F |
| 19 |
|
Revision 1.89 of seaice_growth.F completes the overall october 2010 revision and |
| 20 |
|
merging process. The merged routine has two branches, and one activates the 'evolution' |
| 21 |
|
branch by setting SEAICE_GROWTH_LEGACY to undef (at the top of the routine for now). |
| 22 |
|
In summary, here is what the '10/10 revision' did consist of |
| 23 |
|
part 1: seaice_growth.F revisions 1.71 to 1.75 |
| 24 |
|
- Preliminary seaice_growth.F clean-up, keeping test results strictly unchanged. |
| 25 |
|
The emphasis here was on improving readability (e.g. by trying to use meaningful |
| 26 |
|
variable names) and modularity (e.g. by adding arrays to store heff etc. updates). |
| 27 |
|
part 1.5: |
| 28 |
|
- revising the lab_sea experiment to make it a more useful (i.e. stable |
| 29 |
|
and representative of an actual model run). This was done by generating a |
| 30 |
|
(winter) pickup to start from, switching back the main experiment dynamic |
| 31 |
|
solver from evp to lsr, having one of the experiments use the free drift solution. |
| 32 |
|
part 2: seaice_growth.F revisions 1.76 to 1.85 |
| 33 |
|
- Sorting out hard-coded constants that prevented heat and fresh water conservation. |
| 34 |
|
- Sorting out the various fields units and sign (set early on and reset at the end) |
| 35 |
|
to avoid numerous scalings/re-scalings. By v1.85 the basic unit for ice, heat and |
| 36 |
|
fresh water stock is 'effective ice meters', and >0 implies more ice/snow. |
| 37 |
|
- Using tendency terms arrays (d_HEFF* etc.) to alleviate remaining code locks, |
| 38 |
|
and simplifying logical expressions to further improve readibility. |
| 39 |
|
part 3: seaice_growth.F revisions 1.86 to 1.89 |
| 40 |
|
- Re-organize the blocks of code more logically. |
| 41 |
|
- Merging the '_if.F' pieces that add something new. |
| 42 |
|
o pkg/thsice: |
| 43 |
|
- vectorize thsice_solve4temp and thsice_get_exf. |
| 44 |
|
Problem: thsice_solve4temp has an unvectorizable statement that |
| 45 |
|
might also be problematic for TAF |
| 46 |
|
- thsice_get_exf: add the ALLOW_BULK_LARGEYEAGER04 CPP-flag. This |
| 47 |
|
does not change the verifcation experiment, but will change all |
| 48 |
|
experiments that have this flag undefined (not tested in verification); |
| 49 |
|
a little reorganization to save a few 2D fields and recomputations |
| 50 |
|
- 1st step of revisiting thsice adjoint (major recomp. sorted out) |
| 51 |
|
- 2nd step: fix some store dirs |
| 52 |
|
o seems to work somewhat, i.e. using ifort but not g77 |
| 53 |
|
o likely cause are incomplete init. or recomputation |
| 54 |
|
o start verification/global_ocean.cs32x15/input_ad.thsice |
| 55 |
|
o pkg/obcs: add support for exch2 |
| 56 |
|
- the position of the boundary and prescribed values are specified using |
| 57 |
|
a global domain with exch2 facets stacked |
| 58 |
|
- in x for N,S boundaries (like W2_mapIO=-1) |
| 59 |
|
- in y for E,W boundaries (so E,W boundaries do not overlap) |
| 60 |
|
o pkg/mdsio: add support for writing global sections with exch2 |
| 61 |
|
file layout is as for obcs described above: |
| 62 |
|
- for x-z sections, facets are stacked in x (like W2_mapIO=-1) |
| 63 |
|
- for y-z sections, facets are stacked in y |
| 64 |
|
o pkg/thsice: |
| 65 |
|
- change a few routines for better vectorization: thsice_albedo.F, |
| 66 |
|
thsice_step_temp.F, and in thsice_calc_thickn.F. Results are |
| 67 |
|
not affected, but in particular the latter routine has changed |
| 68 |
|
dramatically (not necessarily easier to read now). |
| 69 |
|
- re-arrange calculation in THSICE_SOLVE4TEMP: move outside of i,j loops |
| 70 |
|
iteration loop and call to S/R THSICE_GET_BULKF & THSICE_GET_EXF. |
| 71 |
|
o verification/lab_sea: |
| 72 |
|
- overhaul of the forward model experiments. The main motivation |
| 73 |
|
for this revision is that lab_sea has proven quite fragile (when |
| 74 |
|
changing the sea-ice code, and switching platforms) to a point |
| 75 |
|
when failures were very hard to attribute to specific code problems. |
| 76 |
|
- the suspected reason for this fragility is noise amplification by evp so, |
| 77 |
|
to start with, we decided to replace evp with lsr in the main experiment. |
| 78 |
|
- the input.lsr experiment thus became obsolete and was removed. |
| 79 |
|
- it was replaced with the input.fd experiment that uses an |
| 80 |
|
analytical "free drift" solution as the ice velocity field. |
| 81 |
|
- Taken together, input (lsr), input.fd (free drift) and input.hb87 (evp) |
| 82 |
|
will hopefully allow us to discriminate problems due to dynamic solvers |
| 83 |
|
from problems due to the thermodynamic code. input.salt_plume (lsr) allows |
| 84 |
|
us to test additional options, along with the salt plume package. |
| 85 |
|
- Finally a pickup was generated (at the end of a one year run) to |
| 86 |
|
make the tests more representative of an actual model run. Previously |
| 87 |
|
the tests starting point was a 1m thickness of ice everywhere. |
| 88 |
|
o pkg/seaice: |
| 89 |
|
- introducing S\R seaice_freedrift.F that compute the |
| 90 |
|
analytical solution to an approximate ice momentum |
| 91 |
|
equation (i.e. the simple stress/coriolis balance). |
| 92 |
|
- define SEAICE_ALLOW_FREEDRIFT to trigger the computation |
| 93 |
|
of the free drift velocity field, and enable its use. |
| 94 |
|
To actually use the free drift solution as the model ice |
| 95 |
|
velocity field set SEAICEuseFREEDRIFT to .TRUE. in data.seaice. |
| 96 |
|
o move "useOASIS" from PARAMS.h to EEPARAMS.h and read it from "eedata" |
| 97 |
|
|
| 98 |
|
checkpoint62l (2010/10/04) |
| 99 |
|
o pkg/seaice: |
| 100 |
|
-fix the USE_ORIGINAL_SBI version (for variable freezing point); |
| 101 |
|
-use "celsius2K" in SEAICE_VARIABLE_FREEZING_POINT and in new |
| 102 |
|
version (with USE_ORIGINAL_SBI undef); |
| 103 |
|
o add hooks for new packages OASIS, the package will follow |
| 104 |
|
(hooks exclude the seaice pkg for now) |
| 105 |
|
o pkg/autodiff: ALLOW_AUTODIFF_WHTAPEIO |
| 106 |
|
Option to use mdsio_read_whalos.F/mdsio_write_whalos.F, |
| 107 |
|
rather than mdsio_readvector.F/mdsio_writevector.F, within |
| 108 |
|
adread_adwrite.F to read/write adjoint-related tapes to disk. |
| 109 |
|
o eesupp and pkg/mdsio: ALLOW_WHIO |
| 110 |
|
Option to write/read 2D fields to files including tiles halos. |
| 111 |
|
The main purpose is for adjoint related "tape I/O". |
| 112 |
|
The secondary purpose is for debugging phases. |
| 113 |
|
A word on those rather specifically intended purposes. |
| 114 |
|
The code was meant to be stripped down to the bare minimum. |
| 115 |
|
Typically, there is a unique and basic mapping to the global |
| 116 |
|
buffer (tile by tile, then proc by proc, then level by level), and |
| 117 |
|
byteswaps and meta-files are omitted. Indeed the (*_WH*) code is not |
| 118 |
|
meant to generate user friendly outputs of various shapes and forms. |
| 119 |
|
Rather it is intended, for the knowledgeabale developers, |
| 120 |
|
to be largely independent of the many mdsio options, to allow |
| 121 |
|
outputs of tiles+halos in debugging phases and in adjoint runs, |
| 122 |
|
and to allow some flexibility in tuning (adoint) I/O depending on each |
| 123 |
|
file system behavior. With regard to tuning, most of the potential knobs |
| 124 |
|
(to leave files open, to allocate the I/O indep. of usesinglecpuio, |
| 125 |
|
to specify directories, to specify the typical size of I/O calls, etc.) |
| 126 |
|
are not yet included, and will be added if they prove useful. |
| 127 |
|
o pkg/seaice: |
| 128 |
|
fix uninitialised arrays in USE_ORIGINAL_SBI version of seaice_solve4temp.F |
| 129 |
|
o pkg/seaice: |
| 130 |
|
new S/R seaice_solve4temp.F: this is the merged version of Ian Fenty |
| 131 |
|
code (seaice_budget_ice_if.F) which enable to recover old standard |
| 132 |
|
code (seaice_budget_ice.F) when option USE_ORIGINAL_SBI is defined. |
| 133 |
|
temporary: add "#define USE_ORIGINAL_SBI" at the top of seaice_solve4temp.F. |
| 134 |
|
o model/src (rStar): |
| 135 |
|
use simple average (not area weighted) of rStarFac at U,V point when using |
| 136 |
|
vectorInvariant and selectKEscheme=1,3 (consistent with mom. vertical advection) |
| 137 |
|
update output of exp. fizhi-cs-32x32x40 and fizhi-cs-aqualev20. |
| 138 |
|
|
| 139 |
|
checkpoint62k (2010/09/17) |
| 140 |
|
o adjoint |
| 141 |
|
* changed verification/global_ocean.cs32x15 ; adjoint now tests |
| 142 |
|
Adams-Bashforth 3 + NLFS + RSTAR |
| 143 |
|
* some more modifs required for ECCO-GODAE v4 |
| 144 |
|
o pkg/seaice |
| 145 |
|
add SEAICE_AGE_VOL option to use volume instead of area for |
| 146 |
|
integrating SEAICE_AGE |
| 147 |
|
o model/src: |
| 148 |
|
first check-in of sigma & hybrid-sigma coordinate code ; also includes |
| 149 |
|
non-linear free-surface parts ; but pressure-gradient not yet coded. |
| 150 |
|
o pkg/ptracers: |
| 151 |
|
option for ptracer to convert Salt Relax into additional EmP. |
| 152 |
|
tested in exp. global_ocean.90x40x15.dwnslp (update output) |
| 153 |
|
|
| 154 |
|
checkpoint62j (2010/08/25) |
| 155 |
|
o pkg/ggl90: |
| 156 |
|
other minor changes: remove 3-D temp array gTKE (+ add argument myIter) |
| 157 |
|
o pkg/dic: car_flux.F, phos_flux.F |
| 158 |
|
- reduce number of exp-calls in order to improve performance |
| 159 |
|
- clean up a little (replace fields by their macros, etc., fix |
| 160 |
|
typo in CPP-flag) |
| 161 |
|
o pkg/ggl90: minor changes |
| 162 |
|
- GGL90diffKrS was removed --> always use GGL90diffKr |
| 163 |
|
- GGL90viscAr was removed --> replaced with GGL90viscArU, GGL90viscArV |
| 164 |
|
- hack of mxlMaxFlag=2 --> ensure mixing between first and second level |
| 165 |
|
(commented out for now) |
| 166 |
|
- change in max/min operations to ensure that smoothing is ok |
| 167 |
|
- smoothing of GGL90viscAr was moved to ggl90_calc.F (as done for GGL90diffKr) |
| 168 |
|
- always use diffKrNrT as background profile (i.e. never use diffKr field) |
| 169 |
|
o pkg/cfc: bug fix, AREA(i,j,1,bi,bj) -> AREA(i,j,bi,bj) |
| 170 |
|
o pkg/dic: |
| 171 |
|
- model/src/impldiff.F: if TARGET_NEC_SX, extend loop ranges for better |
| 172 |
|
vectorization |
| 173 |
|
- small bug fix in insol for the unlikely case of a cartesian grid |
| 174 |
|
- towards vectorizing the dic package: bio_export.F, car_flux.F, phos_flux.F |
| 175 |
|
+ now vectorize, which also removes adjoint recomputations; |
| 176 |
|
+ replace integer maksp1 by _RL maskp1 |
| 177 |
|
+ replace a**x by exp(x*log(a)) in phos_flux.F for speed |
| 178 |
|
|
| 179 |
|
checkpoint62i (2010/07/21) |
| 180 |
|
o pkg/flt: |
| 181 |
|
- with OBCS: stop floats which enter the OB region |
| 182 |
|
o model/src: |
| 183 |
|
- use tempVertAdvScheme & saltVertAdvScheme (instead of horizontal ones) |
| 184 |
|
for implicit vertical advection. |
| 185 |
|
o pkg/seaice: |
| 186 |
|
- account for melting in seaice age evolution. |
| 187 |
|
- remove double counting of seaice ridging effect on seaice age. |
| 188 |
|
|
| 189 |
|
checkpoint62h (2010/06/18) |
| 190 |
|
o pkg/exch2: |
| 191 |
|
- remove exch2_*xy_rx files: call was replaced with exch2_*3d_rx(myNz=1). |
| 192 |
|
o tools/genmake2 |
| 193 |
|
- remove "-pdefault" option: redundant with the use of "packages.conf"; |
| 194 |
|
- remove file pkg/pkg_default : set the default as a pkg group |
| 195 |
|
("default_pkg_list") defined in pkg/pkg_groups; |
| 196 |
|
- fix for case pkg/pkg_depend is missing (was hanging before); |
| 197 |
|
- add option "-pgroups" to specify a different pkg_groups than the default. |
| 198 |
|
o verification/adjustment.cs-32x32x1: |
| 199 |
|
add a "minimal" set-up test (code_min & input_min) which only compiles |
| 200 |
|
eesupp/src and the 2 pkgs (exch2 & debug) which don't need model/src & inc. |
| 201 |
o tools/genmake2 |
o tools/genmake2 |
| 202 |
- extend pkg_depend rules to model & eesupp (e.g., if model depend on [pkg])) |
- extend pkg_depend rules to model & eesupp (e.g., if model depend on [pkg])) |
| 203 |
- fix details of checking for mnc & profile pkg when NetCDF is unavailable. |
- fix details of checking for mnc & profile pkg when NetCDF is unavailable. |
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch