| Version 22 (modified by techene, 3 years ago) (diff) |
|---|
Star coordinate faster implementation
Change the way to deal with the vertical scale factors in NEMO in order to save parallel processing time for z-star coordinate. This modification can be activated through a cpp key : key_qco.
Last edition: Wikinfo(changed_ts)? by Wikinfo(changed_by)?
The PI is responsible to closely follow the progress of the action, and especially to contact NEMO project manager if the delay on preview (or review) are longer than the 2 weeks expected.
Summary
| Action | optimisation of the vertical scale factor e3 computation |
|---|---|
| PI(S) | Techene, Madec |
| Digest | compute e3 on the fly from e3_0(:,:,:,Ktl) * ( 1 + ssh(:,:,Ktl) / h_0( :,: ) * mask( :,:,: ) instead of storing e3t/u/v/w/f... |
| Dependencies | If any |
| Branch | source:/NEMO/branches/2020/dev_r12377_KERNEL-06_techene_e3 |
| Branch | source:/NEMO/branches/2020/dev_r13327_KERNEL-06_2_techene_e3 |
| Previewer(s) | Madec, Chanut, Masson |
| Reviewer(s) | Madec |
| Ticket | #2385 #2527 #2555 |
Description
In z* vertical configuration, NEMO r12377 uses memory to store and update vertical scale factors e3[P] where P = {t-, u-, v-, w-, f-, uw-, vw-} points at "before", "now" and "after" time steps. This means memory storage 6 x 4D + 1 x 3D tables, memory acces and CPU time for updating 3D scale factors. The code modification consists in computing scale factor e3[P] on the fly using each time it is needed with formula e3[P](Kt) = e3[P]_0 * (1 + r3[P](Kt) * mask[P]) where r3P(Kt) (= ssh[P](Kt) / h_0) is a 2D table computed from ssh update at P = {u-,v-,f-} points accordingly with ssh update along a step. This change is only applied in case key_qco is activated.
Because we reduce the number of tables reached in memory we have a better chance to keep using fast RAM memory. Because we no longer compute 3D interpolation but 2D instead algorithm complexity is smaller and use less CPU time. Both make computation about 10% faster whatever the domain size (tested between 10x10 to 100x100 points per computation node). when cutting communications.
This branche also comes with improvements from KERNEL-07 such as symmetric diffusion tensor #2527 implemented in dynldf_lap_blp used controled by nn_dynldf_typ namelist parameter. It contains dynvor correction for using ln_dynvor_msk and a proper fix for using ENS and ENE with partial steps described in #2555. Finally a new shallow water test case has been added.
Implementation
KERNEL-06's version 1 [pre mid-merge 2020] : /NEMO/branches/2020/dev_r12377_KERNEL-06_techene_e3
When key_qco is not activated NEMO should be produce exactly same results as the trunk and passes SETTE version r13167 has been delivered for mid-merge party !
Modification of the code have been implemented step by step. This enabled to stress out some interesting features.
- e3. expression involves tables of distinct dimension then e3.(:,:,: ) call fails it may be necessary to introduce temporary variables (same for water height expression)
- "e3. =" is no longer possible
- e3t/u/v/f modifications did not introduce any difference in the results, e3w modification does because both approaches vvl and qco do not take into account of the bottom level in the same way
- in GYRE e3w_0 are not the half sum of e3u_0 so the way it is implemented in the reference version is not convenient
- e3. substitution makes lines longer than 136 character this may be a problem for compilers (most have been checked but not all)
- ssh filtering has been displaced upper in order to provide filtered r3P in TOP asselin filtering
KERNEL-06's version 2 [post mid-merge 2020] : dev_r13327_KERNEL-06_2_techene_e3
When key_qco is activated NEMO passes suitable SETTE tests i.e. custom GYRE_PISCES, ORCA2_ICE_PISCES, SPITZ12, AMM12. ISOMIP cannot run yet nor AGRIF.
Pseudo merge with NEMO/branches/2020/dev_r12527_Gurvan_ShallowWater consists mainly in modifying dynldf_lap_blp for adding symmetric tensor operator and dynvor for cleaning and diawri for F-point outputs and adding a test case with specific RK3 step.
Documentation updates
Need for documentation on qco changes ! Need for documentation on symetric tensor for viscosity operator Need for documentation on e3f computation in ENE and ENS in zps ? + dynspg_ts !
Preview
...
Review
...
