Substepping outside dycore, refactor grid_wrapper, add mean_cell_area and vol_flx_ic #683
Conversation
model/atmosphere/dycore/src/icon4py/model/atmosphere/dycore/solve_nonhydro.py
Show resolved
Hide resolved
havogt
changed the title
|
cscs-ci run default |
havogt
changed the title
model/atmosphere/dycore/tests/dycore_tests/test_solve_nonhydro.py
Outdated
Show resolved
Hide resolved
| vertical_end=gtx.int32(self._vertical_params.nflat_gradp + 1), | ||
| offset_provider=self._grid.offset_providers, | ||
| ) | ||
| if self._vertical_params.nflatlev < gtx.int32(self._vertical_params.nflat_gradp + 1): |
There was a problem hiding this comment.
Can it happen that nflatlev>= nflat_gradp + 1 ? Shouldn't the condition rather be checked when computing these values at initialization? The other questions is does gt4py crash for a "negative" or "empty" domain?
There was a problem hiding this comment.
Yes, it crashed for an empty domain. We should probably fix that on the GT4Py side.
There was a problem hiding this comment.
Yes, I would suggest so. Anyway, we can leave this as a workaround for now. Can you open an issue in GT4Py?
There was a problem hiding this comment.
On a second thought, I am not sure if it should be legal, because that could also hide user errors. At least GT4Py should warn that it doesn't execute a program. Here is the issue GridTools/gt4py#1891
There was a problem hiding this comment.
Thanks. I agree it could hide errors, but there are valid cases for it: for example if you run something like APE every layer is flat, so if you try to compute something from some layer up to the first flat layer, that might collapse (what probably happens in the example here) or in the horizontal if you compute something only on the boundaries, but you have a global model, then there are no boundaries. The user would then have to always do an extra
if domain-is-not-empty check what the if condition above essentially is.
| @@ -256,6 +245,7 @@ def test_dycore_wrapper_granule_inputs( | |||
|
|
|||
| # PrepAdvection | |||
| vn_traj = sp.vn_traj() | |||
| vol_flx_ic = data_alloc.zero_field(icon_grid, dims.CellDim, dims.KDim) # TODO sp.vol_flx_ic() | |||
There was a problem hiding this comment.
hm, you wanted the vol_flx_ic in the init savepoint? I think I only added it to the exit... shame on me...
|
cscs-ci run default |
|
cscs-ci run default |
|
cscs-ci run default |
|
Mandatory Tests Please make sure you run these tests via comment before you merge!
Optional Tests To run benchmarks you can use:
To run tests and benchmarks with the DaCe backend you can use:
In case your change might affect downstream icon-exclaim, please consider running
For more detailed information please look at CI in the EXCLAIM universe. |
|
tests passed in #687 |
Model:
Fortran wrappers:
max_nudge_coeffto indicate it's the ICON scaled one that is passed to the bindings