Changes between Version 11 and Version 12 of 2020WP/HPC-02_Daley_Tiling

Timestamp:

2020-11-18T18:53:09+01:00 (4 years ago)

Author:

hadcv

Comment:

--

2020WP/HPC-02_Daley_Tiling

@@ -438 +438 @@
 
 == Tests
-
-{{{#!box width=50em info
-[[Include(wiki:Developers/DevProcess#tests)]]
-}}}
-
-''...''
+=== SETTE
+
+SETTE passes all tests (including with tiling turned on) and compares with the trunk.
+The Intel compiler (ifort 18.0.5 20180823) is used with XIOS in detached mode. 
+The Cray compiler was not used due to #2394 (new Cray compiler does not work with new way of reading namelists) and due to older Cray compilers raising various errors.
+
+==== Regular checks
+
+* Can this change be shown to produce expected impact (option activated)? __YES__
+* Can this change be shown to have a null impact (option not activated)? __YES__
+* Results of the required bit comparability tests been run: are there no differences when activating the development? __YES (SETTE), NO (other tests)__
+  * If some differences appear, is reason for the change valid/understood? __YES (see known failures)__
+  * If some differences appear, is the impact as expected on model configurations? __YES__
+* Is this change expected to preserve all diagnostics? __NO (see known failures)__
+  * If no, is reason for the change valid/understood? __YES__
+* Are there significant changes in run time/memory? __NO__
+
+==== Detailed SETTE results
+
+Tiling is turned on in this report.
+
+{{{
+Current code is : NEMO/branches/2020/dev_r13383_HPC-02_Daley_Tiling @ r13745  ( last change @ r13745 )
+
+SETTE validation report generated for : 
+
+       NEMO/branches/2020/dev_r13383_HPC-02_Daley_Tiling @ r13745+ (last changed revision)
+
+       on XC40_METO_IFORT arch file
+
+
+!!---------------1st pass------------------!!
+
+   !----restart----!   
+WGYRE_PISCES_ST              run.stat    restartability  passed :  13745+
+WGYRE_PISCES_ST              tracer.stat restartability  passed :  13745+
+WORCA2_ICE_PISCES_ST         run.stat    restartability  passed :  13745+
+WORCA2_ICE_PISCES_ST         tracer.stat restartability  passed :  13745+
+WORCA2_OFF_PISCES_ST         tracer.stat restartability  passed :  13745+
+WAMM12_ST                    run.stat    restartability  passed :  13745+
+WORCA2_SAS_ICE_ST            run.stat    restartability  passed :  13745+
+WAGRIF_DEMO_ST               run.stat    restartability  passed :  13745+
+WWED025_ST                   run.stat    restartability  passed :  13745+
+WISOMIP+_ST                  run.stat    restartability  passed :  13745+
+WOVERFLOW_ST                 run.stat    restartability  passed :  13745+
+WLOCK_EXCHANGE_ST            run.stat    restartability  passed :  13745+
+WVORTEX_ST                   run.stat    restartability  passed :  13745+
+WICE_AGRIF_ST                run.stat    restartability  passed :  13745+
+
+   !----repro----!   
+WGYRE_PISCES_ST              run.stat    reproducibility passed :  13745+
+WGYRE_PISCES_ST              tracer.stat reproducibility passed :  13745+
+WORCA2_ICE_PISCES_ST         run.stat    reproducibility passed :  13745+
+WORCA2_ICE_PISCES_ST         tracer.stat reproducibility passed :  13745+
+WORCA2_OFF_PISCES_ST         tracer.stat reproducibility passed :  13745+
+WAMM12_ST                    run.stat    reproducibility passed :  13745+
+WORCA2_SAS_ICE_ST            run.stat    reproducibility passed :  13745+
+WORCA2_ICE_OBS_ST            run.stat    reproducibility passed :  13745+
+WAGRIF_DEMO_ST               run.stat    reproducibility passed :  13745+
+WWED025_ST                   run.stat    reproducibility passed :  13745+
+WISOMIP+_ST                  run.stat    reproducibility passed :  13745+
+WVORTEX_ST                   run.stat    reproducibility passed :  13745+
+WICE_AGRIF_ST                run.stat    reproducibility passed :  13745+
+
+   !----agrif check----!   
+ORCA2 AGRIF vs ORCA2 NOAGRIF run.stat    unchanged  -    passed :  13745+ 13745+
+
+   !----result comparison check----!   
+
+check result differences between :
+VALID directory : /home/d00/hadcv/cylc-run/u-bs939/share/sette/output/NEMO_VALIDATION at rev 13745+
+and
+REFERENCE directory : /home/d00/hadcv/cylc-run/u-bs939/share/sette_ref/output/NEMO_VALIDATION at rev 13688
+
+WGYRE_PISCES_ST       run.stat    files are identical  
+WGYRE_PISCES_ST       tracer.stat files are identical  
+WORCA2_ICE_PISCES_ST  run.stat    files are identical  
+WORCA2_ICE_PISCES_ST  tracer.stat files are identical  
+WORCA2_OFF_PISCES_ST  tracer.stat files are identical  
+WAMM12_ST             run.stat    files are identical  
+WORCA2_SAS_ICE_ST     run.stat    files are identical  
+WAGRIF_DEMO_ST        run.stat    files are identical  
+WWED025_ST            run.stat    files are identical  
+WISOMIP+_ST           run.stat    files are identical  
+WVORTEX_ST            run.stat    files are identical  
+WICE_AGRIF_ST         run.stat    files are identical  
+WOVERFLOW_ST          run.stat    files are identical  
+WLOCK_EXCHANGE_ST     run.stat    files are identical
+}}}  
+
+=== Development testing
+
+A configuration based on ORCA2_ICE_PISCES (without `key_si3` or `key_top`) was used to test code modified by the tiling development.
+To facilitate cleaner testing, `ln_trabbc`, `ln_trabbl`, `ln_icebergs`, `ln_rnf`, `ln_ssr`, `ln_tradmp`, `ln_ldfeiv`, `ln_traldf_msc`, `ln_mle`, `ln_zdfddm` and `ln_zdfiwm` were all set to false.
+`ln_qsr_2bd` was used instead of `ln_qsr_rgb`, `nn_fsbc` was set to 1, and `nn_ice` and `nn_fwb` were set to 0.
+
+Simulations of the tiling branch were run for 10 days with 1-day diagnostic output, for all scientific options relevant to the affected code.
+Each simulation was repeated with tiling turned on, using square tile sizes of 5 and 50 (the latter being equivalent to one tile over the full domain).
+Additionally, all simulations including those with tiling were repeated with `nn_hls = 2`.
+
+run.stat and diagnostic output were compared with equivalent simulations of the trunk and 10-day simulations of the tiling branch that were run in two 5-day submissions (i.e. testing for restartability).
+
+Version 8.3.4 of the Cray compiler was used with XIOS 2.5 revision 1565.
+A `jpni = 4`, `jpnj = 8` decomposition was used with 6 XIOS processors.
+
+=== Known failures in development tests__
+
+* `ln_trabbl = .true.` with `nn_bbl_adv > 0` gives different results when using tiling
+    
+  This is due to a change in the order of computation of `pt_rhs` in `tra_bbl_adv` when using the tiling.
+  The loop over i is broken up by the tiles, which causes the up-slope and down-slope contributions to be added in a different order on the intersections between tiles.
+
+* Some trends diagnostics have slightly different values at one point on the northfold in the development branch
+
+  This may be because I removed an `lbc_lnk` from `tra_zdf`, similar to the effect of removing the `lbc_lnk` for `utr_bbl`/`vtr_bbl` in `tra_bbl`.
+  It could be restored if necessary, since we do not use tiling with the trends diagnostics.
+
+* NEMO fails with a `stp_ctl` error when using `ln_traldf_hor = .true.` or `ln_zdfosm = .true.`, in both the development branch and trunk
+
+  I assume this is related to my configuration, as I did not have this issue when testing in GYRE.
+
+=== Untested code
+
+* Code (e.g. `zps_hde_isf`) that requires `ln_isfcav = .true.`
+* Code in `diaptr.F90` for the `uocetr_vsum_cumul` diagnostic (`ptr_ci_2d` subroutine)
+
+  XIOS hangs when trying to output this diagnostic, which may be due to the XIOS library being too old
+
+* Code in `tra_adv` that requires `ln_wave .AND. ln_sdw = .true.`, 
+* Code in `tra_adv` that requires `ln_vvl_ztilde .OR. ln_vvl_layer = .true.`
+* Code in `tra_asm_inc` that requires `ln_asmdin = .true.`
+
+  I did not have the required input file (assim_background_state_DI.nc) for ORCA2, although I was able to make an idealised one for testing in GYRE.
+
+* Code in `tra_asm_inc` that requires `ln_temnofreeze = .true.` or `ln_seaiceinc = .true.`
+
+  These logicals are hard coded as false.
 
 == Review