4 |
Notes on tags used in MITgcmUV |
Notes on tags used in MITgcmUV |
5 |
============================== |
============================== |
6 |
|
|
7 |
|
o pkg/seaice: |
8 |
|
- Rename SEAICE_VECTORIZE_LSR_ZEBRA to SEAICE_LSR_ZEBRA and define it |
9 |
|
in lab_sea forward experiment (changes results -> updated) |
10 |
|
o pkg/seaice: modify seaice_lsr.F in order to improve the gradient |
11 |
|
computations; for SEAICE_VECTORIZE_LSR |
12 |
|
- move the loops over which the tridiagonal solvers (seaice_lsr_tridiagu/v) |
13 |
|
sweep to the outside, |
14 |
|
- remove store directives and add "CADJ loop sequential" directives |
15 |
|
in analogy to model/src/solve_tridiagonal.F |
16 |
|
- replace many "#ifdef SEAICE_VECTORIZE_LSR_ZEBRA" by variable loop |
17 |
|
boundaries to yield more readable code. This has also the charming side |
18 |
|
effect that your can use SEAICE_VECTORIZE_LSR_ZEBRA without |
19 |
|
SEAICE_VECTORIZE_LSR (i.e. adjoint without recomputations in |
20 |
|
seaice_lsr_tridiagu/v now requires either of these flags, vectorization |
21 |
|
still requires SEAICE_VECTORIZE_LSR with SEAICE_VECTORIZE_LSR_ZEBRA as |
22 |
|
an additional option) |
23 |
|
The tridiagonal solvers are now completely analoguous to solve_tridiagonal.F |
24 |
|
but the adjoint gradients (which are affected by this change) still explode. |
25 |
o coupler (atm & ocn _compon_interf): |
o coupler (atm & ocn _compon_interf): |
26 |
- avoid using Exch2 IO mapping if not suitable. |
- avoid using Exch2 IO mapping if not suitable. |
27 |
|
|
28 |
|
>>>>>>> 1.2160 |
29 |
checkpoint65m (2015/06/15) |
checkpoint65m (2015/06/15) |
30 |
o pkg/layers: |
o pkg/layers: |
31 |
- clean-up old parameters setting (from Sep 2012): |
- clean-up old parameters setting (from Sep 2012): |