Opened 3 years ago
Last modified 2 years ago
#2605 new Task
KNL-01_Sibylle_RK3_stage1
| Reported by: | techene | Owned by: | techene |
|---|---|---|---|
| Priority: | high | Milestone: | Unscheduled |
| Component: | MULTIPLE | Version: | trunk |
| Severity: | minor | Keywords: | time-stepping, RK3 |
| Cc: |
Description
Workplan action
Wikipage: wiki:2021WP/KNL-01_Sibylle_RK3_stage1
Commit History (43)
| Changeset | Author | Time | ChangeLog |
|---|---|---|---|
| 15630 | techene | 2022-01-05T19:02:55+01:00 | #2605 correct r3f bug in RK3 |
| 15628 | techene | 2022-01-04T22:41:46+01:00 | #2605 correct reproducibility bug for MLF introduced in r15626 |
| 15627 | techene | 2022-01-04T19:30:17+01:00 | #2605 qco r3. ratios management optimised (RK3) and some cleanning (MLF) |
| 15626 | techene | 2022-01-04T18:24:17+01:00 | #2605 VORTEX should pass sette TBC |
| 15625 | techene | 2022-01-04T16:39:00+01:00 | #2605 ISOMIP+ now passes sette |
| 15623 | techene | 2021-12-30T10:04:54+01:00 | #2605 RK3 sette (except VORTEX) successful with RK3 (ORCA2_ICE_PISCES, AMM12, LOCK_EXCHANGE, OVERFLOW, WED025) in DEBUG mode (rhop is required for tramle and some dia routines, uu/vv/ts@Kmm needs to be initialised for routines such as ssm_sbc_init) |
| 15622 | techene | 2021-12-29T10:52:29+01:00 | #2605 Kmm not used for tke initialization |
| 15621 | techene | 2021-12-29T10:51:58+01:00 | #2605 update RK3 loops and add lbc_lnk accordinly |
| 15620 | techene | 2021-12-29T10:51:43+01:00 | #2605 update RK3 loops according to the trunk transformation |
| 15592 | techene | 2021-12-13T11:30:55+01:00 | #2605 : remove obsolete lbc_lnk_multi |
| 15586 | techene | 2021-12-08T12:16:15+01:00 | #2605 #2715 bug fix sette without tiling in debug mode (sette.sh -At) now OK except ICE_AGRIF |
| 15581 | techene | 2021-12-07T14:08:22+01:00 | #2605 #2715 bug fix for AGRIF : sette OK without tiling (sette.sh -At) |
| 15574 | techene | 2021-12-03T20:32:50+01:00 | |
| 15561 | techene | 2021-11-30T14:25:02+01:00 | |
| 15532 | techene | 2021-11-24T12:47:32+01:00 | |
| 15514 | techene | 2021-11-16T09:58:22+01:00 | |
| 15513 | techene | 2021-11-15T18:31:29+01:00 | #2605 RK3 : correct dia_ptr initialisation and add a switch for diags/trends logicals |
| 15512 | techene | 2021-11-15T18:22:03+01:00 | #2605 RK3 : add logicals to deactivate trends, dianogstics and outputs at stages 1 & 2 |
| 15321 | techene | 2021-10-04T12:24:15+02:00 | |
| 15083 | techene | 2021-07-05T18:29:04+02:00 | #2605 RGBc debug : ORCA2 tests are now reproducible |
| 15028 | techene | 2021-06-19T10:53:10+02:00 | #2605 : additionnal cleaning sette ~OK (as stated previously) for MLF and RK3 |
| 14993 | techene | 2021-06-15T00:35:18+02:00 | #2605 adapt and optimize mass flux forcing for RK3 |
| 14992 | techene | 2021-06-14T22:26:12+02:00 | #2605 : debug restartability in debug mode |
| 14991 | techene | 2021-06-14T21:52:31+02:00 | #2605 uu_b@Kmm is not used in RK3, in addition it prevents from a single float exception error in debug mode when restarting since Kmm values are not initialised |
| 14990 | techene | 2021-06-14T21:19:43+02:00 | #2605 : new optimisation of traqsr |
| 14949 | acc | 2021-06-04T17:12:54+02:00 | #2605 : bug fixes to allow all SETTE tests to compile and run on this branch without key_RK3. This is a missed variable name change in stp2d.F90 and an uninecessarily duplicated routine in stprk3.F90. The duplicated routine (finalize_lbc) is unused but AGRIF compilations will pick up the incompatible first version from stpmlf.F90 (AGRIF does not respect scope). The copy in stprk3.F90 has been removed. |
| 14943 | acc | 2021-06-03T16:53:29+02:00 | #2605 : Corrections to istate.F90 and trasbc.F90 so that changes to ensure restartability with key_RK3 do not break restartability without key_RK3. Both now passing SETTE tests. |
| 14942 | acc | 2021-06-03T15:56:32+02:00 | #2605 : Tidier version of restart.F90 that avoids writing the newly-introduced, before barotropic velocities when key_RK3 is not active |
| 14941 | acc | 2021-06-03T13:42:27+02:00 | #2605 : restartability and reproducibility fixes (at least for AMM12 with key_qco and key_RK3). Including: Sibylles corrected indexing of ssh in stprk3.F90; inclusion of before barotropic velocities in restarts (restart.F90 and istate.F90) and (as yet, unreviewed) changes to stprk3_stg.F90/trasbc.F90 to ensure correct stage calculations. A couple of other minor changes in module_example.F90 and mppini.F90 that the compiler flagged. |
| 14808 | jchanut | 2021-05-07T14:00:45+02:00 | #2605: reproducibility issue with RK3: hdiv |
| 14800 | jchanut | 2021-05-06T17:42:46+02:00 | #2605: Main changes for a straightforward use of AGRIF with RK3 |
| 14799 | jchanut | 2021-05-06T15:57:06+02:00 | #2605: Correct Vortex setup by reverting some changes introduced @r14223 |
| 14784 | techene | 2021-05-04T21:08:42+02:00 | #2605 : enable diahsb and fix rnf forcing accordingly |
| 14782 | acc | 2021-05-04T17:15:21+02:00 | Branch dev_r14318_RK3_stage1: minor changes to fix compilation errors with VORTEX test case. stp2d.F90 and stprk3_stg.F90 only. #2605 |
| 14604 | techene | 2021-03-09T14:12:13+01:00 | #2605 : allow time filtering of barotropic variables |
| 14587 | techene | 2021-03-04T18:51:35+01:00 | #2605 : cosmetic only |
| 14549 | techene | 2021-02-26T11:03:18+01:00 | #2605 small bug correction in ln_linssh=T |
| 14548 | techene | 2021-02-25T18:16:27+01:00 | #2605 small bug correction |
| 14547 | techene | 2021-02-25T18:07:15+01:00 | #2605 RK3 time-stepping on OCE only (run on GYRE_PISCES without key_top) |
| 14428 | techene | 2021-02-10T19:12:36+01:00 | #2605 loop optimisation from gm |
| 14418 | techene | 2021-02-08T21:55:49+01:00 | #2605 correct the truncature change of ww induced by diawri when ln_zad_Aimp=T along a time step |
| 14381 | techene | 2021-02-03T13:36:25+01:00 | #2605 : add advection velocity as input in advective flux form routines |
| 14353 | techene | 2021-01-27T15:57:05+01:00 | #2605 branch for RK3 time-stepping |
Change History (54)
comment:1 Changed 3 years ago by techene
comment:2 Changed 3 years ago by techene
In 14381:
comment:3 Changed 3 years ago by techene
In 14418:
comment:4 Changed 3 years ago by techene
In 14428:
comment:5 Changed 3 years ago by techene
In 14547:
comment:6 Changed 3 years ago by techene
In 14548:
comment:7 Changed 3 years ago by techene
In 14549:
comment:8 Changed 3 years ago by techene
In 14587:
comment:9 Changed 3 years ago by techene
In 14604:
comment:10 Changed 3 years ago by acc
In 14782:
comment:11 Changed 3 years ago by techene
In 14784:
comment:12 Changed 3 years ago by jchanut
In 14799:
comment:13 Changed 3 years ago by jchanut
In 14800:
comment:14 Changed 3 years ago by acc
SETTE update: After disappearing down a few rabbit-holes, I've finally traced the reproducibility problems to inconsistent ww values in the halo at RK3 stage3. The halo value is used in dyn_zad when averaging to uw-, vw- points; and this leads to the differences. My, simple-minded solution is to add an lbc_lnk:
-
stprk3_stg.F90
165 165 ! 166 166 !===>>>>>> Modify dyn_adv_... dyn_keg routines so that Krhs to zero useless 167 167 ! ! advection (VF or FF) ==> RHS 168 CALL lbc_lnk_multi( 'stp_RK3_stg', ww(:,:,:), 'T', 1. ) 168 169 CALL dyn_adv( kstp, Kbb, Kmm , uu, vv, Krhs, zaU, zaV, ww ) 169 170 ! ! Coriolis / vorticity ==> RHS 170 171 CALL dyn_vor( kstp, Kmm , uu, vv, Krhs )
but there may be other solutions (and there may be more to do if ln_zad_Aimp is active). This fixes reproducibility for AMM12 and VORTEX (which are the only two, currently active tests). Interestingly, it also fixes restartability for LOCK_EXCHANGE; but not the others:
Using DEUG settings !----restart----! WAMM12_ST run.stat restartability FAILED : 14784+ (results are different after 7 time steps) WOVERFLOW_ST run.stat restartability FAILED : 14784+ (results are different after 8 time steps) WLOCK_EXCHANGE_ST run.stat restartability passed : 14784+ WVORTEX_ST run.stat restartability FAILED : 14784+ (results are different after 7 time steps) !----repro----! WAMM12_ST run.stat reproducibility passed : 14784+ WVORTEX_ST run.stat reproducibility passed : 14784+
I'll move on to looking at restartability and won't submit a change until alternative solutions have been ruled out.
comment:15 Changed 3 years ago by jchanut
In 14808:
comment:16 Changed 3 years ago by acc
SETTE update: Confirmation that changeset:14808 more elegantly fixes the reproducibility error described in comment 15:
Using DEBUG settings !----restart----! WAMM12_ST run.stat restartability FAILED : 14808 (results are different after 7 time steps) WOVERFLOW_ST run.stat restartability FAILED : 14808 (results are different after 8 time steps) WLOCK_EXCHANGE_ST run.stat restartability passed : 14808 WVORTEX_ST run.stat restartability FAILED : 14808 (results are different after 7 time steps) !----repro----! WAMM12_ST run.stat reproducibility passed : 14808 WVORTEX_ST run.stat reproducibility passed : 14808
comment:17 Changed 3 years ago by acc
SETTE update: No answer to the restartability issue yet but I'm puzzled by the dom_qco initialisation. AMM12_ST/SHORT/ocean.output, for example, reports:
rst_read_ssh : ssh initialization
~~~~~~~~~~~~
Kbb sea surface height read in the restart file
read sshb (rec: 1) in ./AMM12_LONG_00000006_restart_0000.nc ok
dom_qco_init : Variable volume activated
~~~~~~~~~~~~
but dom_qco_init needs sshb and sshn to set all the r3 fields. I'm puzzled because attempts to fix this by including sshn in the restart don't seem to make any difference.
comment:18 Changed 3 years ago by acc
SETTE update Sibylle confirms that only sshb/r3t(:,:,Kbb) is required for RK3; the initialisation of r3t(:,:,Kmm) in dom_qco_init is to support MLF and is redundant/harmless for RK3. Further tracking of restart issues suggests there are compounded causes:
- First up is a small difference in uu_b and vv_b between continuous and restarted runs at the start of the restart step. All inputs to the calculation match between the restarted job and the continuous case at the same stage. Possibly this is just a result of the difference in the order of numerical operations? Adding uu_b and vv_b to the restart and disabling their recalculation in istate.F90, removes this difference. The restart error is slightly reduced but not by much.
- The next discovered difference is in the ts trends after tra_sbc at stage 2. tra_sbc does very little at stage 2. As a test, disabling the updates that use sbc_tsc and rnf_tsc removes any difference. Deciding what is the correct solution is trickier. Note, though that there may be some mixup with the correct before flux for the restart since values appear to be written/overwritten at each stage:
grep sbc_sc_b ocean.output
read sbc_sc_b (rec: 1) in ./AMM12_LONG_00000006_restart_0000.nc ok
read sbc_sc_b (rec: 1) in ./AMM12_LONG_00000006_restart_0000.nc ok
read sbc_sc_b (rec: 1) in ./AMM12_LONG_00000006_restart_0000.nc ok
iom_nf90_rp0123d, file: ./AMM12_SHORT_00000012_restart_0000.nc, var: sbc_sc_b defined ok
iom_nf90_rp0123d, file: ./AMM12_SHORT_00000012_restart_0000.nc, var: sbc_sc_b written ok
iom_nf90_rp0123d, file: ./AMM12_SHORT_00000012_restart_0000.nc, var: sbc_sc_b written ok
iom_nf90_rp0123d, file: ./AMM12_SHORT_00000012_restart_0000.nc, var: sbc_sc_b written ok
- Unfortunately, restartability differences persist even after temporarily. by-passing tra_sbc generated differences.
Investigations continue...
comment:19 Changed 3 years ago by acc
SETTE update Ha, turns out sshn IS required. The final difference (detectable in tracer trends after tra_ldf) turns out to be due to slope differences traced back to ldf_slp and eos_rab. This is because buried in these routines is the use of gdept(:,:,:,:,Kmm). This dependency could be removed for eos_rab and bn2 by using Nbb as the last argument instead of Nnn. However, the use of Kmm is more deeply ingrained in ldf_slp (possibly a trunk-based bug?) and the consequences of changing this are less obvious. I leave this to the real experts. Meanwhile, here are the uncommitted changes that can restore restartability:
- Additions to restart (need to revisit the uu_b, vv_b case to see if the small differences were significant now the larger errors have been found)
-
DOM/istate.F90
148 148 ! Do it whatever the free surface method, these arrays being eventually used 149 149 ! 150 150 uu_b(:,:,Kmm) = 0._wp ; vv_b(:,:,Kmm) = 0._wp 151 uu_b(:,:,Kbb) = 0._wp ; vv_b(:,:,Kbb) = 0._wp151 !!uu_b(:,:,Kbb) = 0._wp ; vv_b(:,:,Kbb) = 0._wp 152 152 ! 153 153 !!gm the use of umsak & vmask is not necessary below as uu(:,:,:,Kmm), vv(:,:,:,Kmm), uu(:,:,:,Kbb), vv(:,:,:,Kbb) are always masked 154 154 DO_3D( 1, 1, 1, 1, 1, jpkm1 ) … … 155 155 uu_b(ji,jj,Kmm) = uu_b(ji,jj,Kmm) + e3u(ji,jj,jk,Kmm) * uu(ji,jj,jk,Kmm) * umask(ji,jj,jk) 156 156 vv_b(ji,jj,Kmm) = vv_b(ji,jj,Kmm) + e3v(ji,jj,jk,Kmm) * vv(ji,jj,jk,Kmm) * vmask(ji,jj,jk) 157 157 ! 158 uu_b(ji,jj,Kbb) = uu_b(ji,jj,Kbb) + e3u(ji,jj,jk,Kbb) * uu(ji,jj,jk,Kbb) * umask(ji,jj,jk)159 vv_b(ji,jj,Kbb) = vv_b(ji,jj,Kbb) + e3v(ji,jj,jk,Kbb) * vv(ji,jj,jk,Kbb) * vmask(ji,jj,jk)158 !!uu_b(ji,jj,Kbb) = uu_b(ji,jj,Kbb) + e3u(ji,jj,jk,Kbb) * uu(ji,jj,jk,Kbb) * umask(ji,jj,jk) 159 !!vv_b(ji,jj,Kbb) = vv_b(ji,jj,Kbb) + e3v(ji,jj,jk,Kbb) * vv(ji,jj,jk,Kbb) * vmask(ji,jj,jk) 160 160 END_3D 161 161 ! 162 162 uu_b(:,:,Kmm) = uu_b(:,:,Kmm) * r1_hu(:,:,Kmm) 163 163 vv_b(:,:,Kmm) = vv_b(:,:,Kmm) * r1_hv(:,:,Kmm) 164 164 ! 165 uu_b(:,:,Kbb) = uu_b(:,:,Kbb) * r1_hu(:,:,Kbb)166 vv_b(:,:,Kbb) = vv_b(:,:,Kbb) * r1_hv(:,:,Kbb)165 !!uu_b(:,:,Kbb) = uu_b(:,:,Kbb) * r1_hu(:,:,Kbb) 166 !!vv_b(:,:,Kbb) = vv_b(:,:,Kbb) * r1_hv(:,:,Kbb) 167 167 ! 168 168 END SUBROUTINE istate_init
-
IOM/restart.F90
162 162 ! 163 163 IF( .NOT.ln_diurnal_only ) THEN 164 164 CALL iom_rstput( kt, nitrst, numrow, 'sshb' , ssh(:,: ,Kbb) ) ! before fields 165 CALL iom_rstput( kt, nitrst, numrow, 'sshn' , ssh(:,: ,Kmm) ) ! before fields 165 166 CALL iom_rstput( kt, nitrst, numrow, 'ub' , uu(:,:,: ,Kbb) ) 166 167 CALL iom_rstput( kt, nitrst, numrow, 'vb' , vv(:,:,: ,Kbb) ) 168 CALL iom_rstput( kt, nitrst, numrow, 'uu_b' , uu_b(:,: ,Kbb) ) ! before fields 169 CALL iom_rstput( kt, nitrst, numrow, 'vv_b' , vv_b(:,: ,Kbb) ) ! before fields 167 170 IF( .NOT.lk_SWE ) THEN 168 171 CALL iom_rstput( kt, nitrst, numrow, 'tb' , ts(:,:,:,jp_tem,Kbb) ) 169 172 CALL iom_rstput( kt, nitrst, numrow, 'sb' , ts(:,:,:,jp_sal,Kbb) ) … … 287 290 IF(lwp) WRITE(numout,*) ' Kbb u, v and T-S fields read in the restart file' 288 291 CALL iom_get( numror, jpdom_auto, 'ub' , uu(:,:,: ,Kbb), cd_type = 'U', psgn = -1._wp ) 289 292 CALL iom_get( numror, jpdom_auto, 'vb' , vv(:,:,: ,Kbb), cd_type = 'V', psgn = -1._wp ) 293 CALL iom_get( numror, jpdom_auto, 'uu_b' , uu_b(:,: ,Kbb), cd_type = 'U', psgn = -1._wp ) 294 CALL iom_get( numror, jpdom_auto, 'vv_b' , vv_b(:,: ,Kbb), cd_type = 'V', psgn = -1._wp ) 295 CALL iom_get( numror, jpdom_auto, 'vb' , vv(:,:,: ,Kbb), cd_type = 'V', psgn = -1._wp ) 290 296 IF( .NOT.lk_SWE ) THEN 291 297 CALL iom_get( numror, jpdom_auto, 'tb', ts(:,:,:,jp_tem,Kbb) ) 292 298 CALL iom_get( numror, jpdom_auto, 'sb', ts(:,:,:,jp_sal,Kbb) ) … … 377 383 CALL iom_get( numror, jpdom_auto, 'sshb' , ssh(:,:,Kbb) ) 378 384 ! 379 385 ! !* RK3: Set ssh at Kmm for AGRIF 380 ssh(:,:,Kmm) = 0._wp 386 !ssh(:,:,Kmm) = 0._wp 387 CALL iom_get( numror, jpdom_auto, 'sshn' , ssh(:,:,Kmm) ) 381 388 #else 382 389 ! !* MLF: Read ssh at Kmm 383 390 IF(lwp) WRITE(numout,*)
-
- Temporary by-pass of tra_sbc induced differences (NOT A PROPER SOLUTION)
-
TRA/trasbc.F90
161 161 ! 162 162 DO jn = 1, jpts !== update tracer trend ==! 163 163 DO_2D( 0, 0, 0, 0 ) 164 165 164 !! pts(ji,jj,1,jn,Krhs) = pts(ji,jj,1,jn,Krhs) + zfact * ( sbc_tsc_b(ji,jj,jn) + sbc_tsc(ji,jj,jn) ) & 165 !! & / e3t(ji,jj,1,Kmm) 166 166 END_2D 167 167 END DO 168 168 ! … … 185 185 #if defined key_RK3 186 186 zdep = 1._wp / h_rnf(ji,jj) 187 187 DO jk = 1, nk_rnf(ji,jj) 188 189 188 !! pts(ji,jj,jk,jp_tem,Krhs) = pts(ji,jj,jk,jp_tem,Krhs) + rnf_tsc(ji,jj,jp_tem) * zdep 189 !! IF( ln_rnf_sal ) pts(ji,jj,jk,jp_sal,Krhs) = pts(ji,jj,jk,jp_sal,Krhs) + rnf_tsc(ji,jj,jp_sal) * zdep 190 190 END DO 191 191 192 192 #else
-
With these changes I can achieve:
./sette.sh -Q -t "AMM12 OVERFLOW VORTEX LOCK_EXCHANGE" !----restart----! WAMM12_ST run.stat restartability passed : 14808+ WOVERFLOW_ST run.stat restartability passed : 14808+ WLOCK_EXCHANGE_ST run.stat restartability passed : 14808+ WVORTEX_ST run.stat restartability FAILED : 14808+ !----repro----! WAMM12_ST run.stat reproducibility passed : 14808+ WVORTEX_ST run.stat reproducibility passed : 14808+
Just VORTEX holding out now :)
comment:20 Changed 3 years ago by acc
SETTE update: towards a better solution for tra_sbc.
Firstly, a quick check has confirmed that uu_b and vv_b do need to be passed via the restart to ensure restartability.
Secondly, a closer look at tra_sbc suggests several issues:
- There is no issue with the rnf terms only the sbc_tsc terms
- stprk3_stg does not pass the kstg optional argument so the RK3 logic is not activated
- Correcting this is not enough because zfact is then inconsistent between continuous and restarted runs. In fact it is undefined for stage 2. Also the before value of sbc_tsc is written to the restart at the wrong point.
These changes (as an alternative to step 2 in comment 20) recover restartability with an active tra_sbc; but this may not be what is intended for zfact:
-
stprk3_stg.F90
212 212 213 213 !===>>>>>> stg1&2: Verify the necessity of these trends (we may need it as there are in the RHS of dynspg_ts ?) 214 214 !!gm ====>>>> needed for heat and salt fluxes associated with mass/volume flux 215 CALL tra_sbc( kstp, Kmm, ts, Krhs ) ! surface boundary condition215 CALL tra_sbc( kstp, Kmm, ts, Krhs, kstg ) ! surface boundary condition 216 216 217 217 IF( ln_isf ) CALL tra_isf( kstp, Kmm, ts, Krhs ) ! ice shelf heat flux 218 218 IF( ln_traqsr ) CALL tra_qsr( kstp, Kmm, ts, Krhs ) ! penetrative solar radiation qsr
-
TRA/trasbc.F90
121 121 ! EMP, SFX and QNS effects 122 122 !---------------------------------------- 123 123 ! !== Set before sbc tracer content fields ==! 124 zfact = 0.5_wp 124 125 IF( kt == nit000 .AND. istg_1 == 1 ) THEN !* 1st time-step 125 126 IF( ln_rstart .AND. .NOT.l_1st_euler ) THEN ! Restart: read in restart file 126 zfact = 0.5_wp127 127 IF( ntile == 0 .OR. ntile == 1 ) THEN ! Do only on the first tile 128 128 IF(lwp) WRITE(numout,*) ' nit000-1 sbc tracer content field read in the restart file' 129 129 sbc_tsc(:,:,:) = 0._wp … … 142 142 DO_2D( isi, iei, isj, iej ) 143 143 sbc_tsc_b(ji,jj,:) = sbc_tsc(ji,jj,:) 144 144 END_2D 145 IF( ntile == 0 .OR. ntile == nijtile ) THEN ! Do only on the last tile 146 IF( lrst_oce ) THEN !== write sbc_tsc in the ocean restart file ==! 147 CALL iom_rstput( kt, nitrst, numrow, 'sbc_hc_b', sbc_tsc(:,:,jp_tem) ) 148 CALL iom_rstput( kt, nitrst, numrow, 'sbc_sc_b', sbc_tsc(:,:,jp_sal) ) 149 ENDIF 150 ENDIF 145 151 ENDIF 146 152 ! !== Now sbc tracer content fields ==! 147 153 DO_2D( isi, iei, isj, iej ) … … 166 172 END_2D 167 173 END DO 168 174 ! 169 IF( ntile == 0 .OR. ntile == nijtile ) THEN ! Do only on the last tile170 IF( lrst_oce ) THEN !== write sbc_tsc in the ocean restart file ==!171 CALL iom_rstput( kt, nitrst, numrow, 'sbc_hc_b', sbc_tsc(:,:,jp_tem) )172 CALL iom_rstput( kt, nitrst, numrow, 'sbc_sc_b', sbc_tsc(:,:,jp_sal) )173 ENDIF174 ENDIF175 175 ! 176 176 !---------------------------------------- 177 177 ! River Runoff effects
comment:21 Changed 3 years ago by acc
In 14941:
comment:22 Changed 3 years ago by acc
In 14942:
comment:23 Changed 3 years ago by acc
In 14943:
comment:24 Changed 3 years ago by acc
In 14949:
comment:25 Changed 3 years ago by techene
In 14990:
comment:26 Changed 3 years ago by techene
In 14991:
comment:27 Changed 3 years ago by techene
In 14992:
comment:28 Changed 3 years ago by techene
In 14993:
comment:29 Changed 3 years ago by techene
Commit 15027 ensure restartability OK reproducibility ~OK (when not using RGBc) for MLF, restartability OK reproducibility OK for RK3
Sending src/OCE/IOM/restart.F90
Sending src/OCE/TRA/traqsr.F90
Sending tests/ISOMIP+/MY_SRC/eosbn2.F90
With key_qco only : note that REPRO tests fail when using traqr with RGBc option (following results show a modified version of ORCA2_ICE_PISCES in order to run with RGB only)
!!---------------1st pass------------------!! !----restart----! WGYRE_PISCES_ST run.stat restartability passed : 14993+ WGYRE_PISCES_ST tracer.stat restartability passed : 14993+ WORCA2_ICE_PISCES_ST run.stat restartability passed : 14993+ WORCA2_ICE_PISCES_ST tracer.stat restartability passed : 14993+ WORCA2_OFF_PISCES_ST tracer.stat restartability passed : 14993+ WAMM12_ST run.stat restartability passed : 14993+ WORCA2_SAS_ICE_ST run.stat restartability passed : 14993+ WAGRIF_DEMO_ST run.stat restartability passed : 14993+ WWED025_ST run.stat restartability passed : 14993+ WISOMIP+_ST run.stat restartability passed : 14993+ WOVERFLOW_ST run.stat restartability passed : 14993+ WLOCK_EXCHANGE_ST run.stat restartability passed : 14993+ WVORTEX_ST run.stat restartability passed : 14993+ WICE_AGRIF_ST run.stat restartability passed : 14993+ WSWG_ST run.stat restartability passed : 14993+ !----repro----! WGYRE_PISCES_ST run.stat reproducibility passed : 14993+ WGYRE_PISCES_ST tracer.stat reproducibility passed : 14993+ WORCA2_ICE_PISCES_ST run.stat reproducibility passed : 14993+ WORCA2_ICE_PISCES_ST tracer.stat reproducibility passed : 14993+ WORCA2_OFF_PISCES_ST tracer.stat reproducibility passed : 14993+ WAMM12_ST run.stat reproducibility passed : 14993+ WORCA2_SAS_ICE_ST run.stat reproducibility passed : 14993+ WORCA2_ICE_OBS_ST run.stat reproducibility FAILED : 14993+ (results are different after 9 time steps) WAGRIF_DEMO_ST run.stat reproducibility FAILED : 14993+ (results are different after 7 time steps) WWED025_ST run.stat reproducibility FAILED : 14993+ (results are different after 7 time steps) WISOMIP+_ST run.stat reproducibility passed : 14993+ WVORTEX_ST run.stat reproducibility passed : 14993+ WICE_AGRIF_ST run.stat reproducibility passed : 14993+ WSWG_ST run.stat reproducibility passed : 14993+ !----agrif check----! ORCA2 AGRIF vs ORCA2 NOAGRIF run.stat unchanged - passed : 14993+ 14993+
With key_RK3 and key_qco : note that in this case GYRE_PISCES and ORCA2_ICE_PISCES have been modified in order to test RK3 scheme with no biogechemistry and with a quasi-eulerian coordinate.
!!---------------1st pass------------------!! !----restart----! WGYRE_PISCES_ST run.stat restartability passed : 14949+ WORCA2_ICE_PISCES_ST ocean.output MISSING : 14949+ WORCA2_ICE_PISCES_ST incomplete test WORCA2_OFF_PISCES_ST directory MISSING : 14949+ WAMM12_ST run.stat restartability passed : 14949+ WORCA2_SAS_ICE_ST directory MISSING : 14949+ WAGRIF_DEMO_ST directory MISSING : 14949+ WWED025_ST directory MISSING : 14949+ WISOMIP+_ST directory MISSING : 14949+ WOVERFLOW_ST run.stat restartability passed : 14949+ WLOCK_EXCHANGE_ST run.stat restartability passed : 14949+ WVORTEX_ST run.stat restartability passed : 14949+ WICE_AGRIF_ST directory MISSING : 14949+ WSWG_ST directory MISSING : 14949+ !----repro----! WGYRE_PISCES_ST run.stat reproducibility passed : 14949+ WORCA2_ICE_PISCES_ST ocean.output MISSING : 14949+ WORCA2_ICE_PISCES_ST incomplete test WORCA2_OFF_PISCES_ST directory MISSING : 14949+ WAMM12_ST run.stat reproducibility passed : 14949+ WORCA2_SAS_ICE_ST directory MISSING : 14949+ WORCA2_ICE_OBS_ST directory MISSING : 14949+ WAGRIF_DEMO_ST directory MISSING : 14949+ WWED025_ST directory MISSING : 14949+ WISOMIP+_ST directory MISSING : 14949+ WVORTEX_ST run.stat reproducibility passed : 14949+ WICE_AGRIF_ST directory MISSING : 14949+ WSWG_ST directory MISSING : 14949+
comment:30 Changed 3 years ago by techene
In 15028:
comment:31 Changed 3 years ago by techene
In 15083:
comment:32 Changed 3 years ago by techene
In 15321:
comment:33 Changed 3 years ago by techene
In 15512:
comment:34 Changed 3 years ago by techene
In 15513:
comment:35 Changed 3 years ago by techene
In 15514:
comment:36 Changed 3 years ago by techene
In 15532:
comment:37 Changed 3 years ago by techene
In 15561:
comment:38 Changed 3 years ago by techene
In 15574:
comment:39 Changed 3 years ago by techene
MLF+QCO sette results after the merge
-bash-4.2$ ./sette_rpt.sh
Current code is : NEMO/branches/2021/dev_r14318_RK3_stage1 @ r15576 ( last change @ r15574 )
SETTE validation report generated for :
NEMO/branches/2021/dev_r14318_RK3_stage1 @ r15574 (last changed revision)
on X64_IRENE arch file
!!---------------1st pass------------------!!
!----restart----!
GYRE_PISCES run.stat restartability passed : 15574
GYRE_PISCES tracer.stat restartability passed : 15574
ORCA2_ICE_PISCES run.stat restartability passed : 15574
ORCA2_ICE_PISCES tracer.stat restartability passed : 15574
ORCA2_OFF_PISCES tracer.stat restartability passed : 15574
AMM12 run.stat restartability passed : 15574
ORCA2_SAS_ICE run.stat restartability passed : 15574
AGRIF_DEMO run.stat restartability passed : 15574
AGRIF_DEMO tracer.stat restartability FAILED : 15574 (results are different after 12 time steps)
WED025 run.stat restartability passed : 15574
ISOMIP+ run.stat restartability passed : 15574
OVERFLOW run.stat restartability passed : 15574
LOCK_EXCHANGE run.stat restartability passed : 15574
VORTEX run.stat restartability FAILED : 15574 (results are different after 121 time steps)
ICE_AGRIF ocean.output MISSING : 15574
ICE_AGRIF incomplete test
SWG run.stat restartability passed : 15574
!----repro----!
GYRE_PISCES run.stat reproducibility passed : 15574
GYRE_PISCES tracer.stat reproducibility passed : 15574
ORCA2_ICE_PISCES run.stat reproducibility passed : 15574
ORCA2_ICE_PISCES tracer.stat reproducibility passed : 15574
ORCA2_OFF_PISCES tracer.stat reproducibility passed : 15574
AMM12 run.stat reproducibility passed : 15574
ORCA2_SAS_ICE run.stat reproducibility passed : 15574
ORCA2_ICE_OBS run.stat reproducibility passed : 15574
AGRIF_DEMO run.stat reproducibility passed : 15574
AGRIF_DEMO tracer.stat reproducibility passed : 15574
WED025 run.stat reproducibility passed : 15574
ISOMIP+ run.stat reproducibility passed : 15574
VORTEX run.stat reproducibility passed : 15574
ICE_AGRIF ocean.output MISSING : 15574
ICE_AGRIF incomplete test
SWG run.stat reproducibility passed : 15574
!----agrif check----!
ORCA2 AGRIF vs ORCA2 NOAGRIF run.stat unchanged - passed : 15574 15574
comment:40 Changed 3 years ago by techene
./sette.sh -Atq
MLF+VVL sette results after the merge
-bash-4.2$ ./sette_rpt.sh
Current code is : NEMO/branches/2021/dev_r14318_RK3_stage1 @ r15576 ( last change @ r15574 )
SETTE validation report generated for :
NEMO/branches/2021/dev_r14318_RK3_stage1 @ r15574 (last changed revision)
on X64_IRENE arch file
!!---------------1st pass------------------!!
!----restart----!
GYRE_PISCES run.stat restartability passed : 15574
GYRE_PISCES tracer.stat restartability passed : 15574
ORCA2_ICE_PISCES run.stat restartability passed : 15574
ORCA2_ICE_PISCES tracer.stat restartability passed : 15574
ORCA2_OFF_PISCES tracer.stat restartability passed : 15574
AMM12 ocean.output MISSING : 15574
AMM12 incomplete test
ORCA2_SAS_ICE run.stat restartability passed : 15574
AGRIF_DEMO run.stat restartability passed : 15574
AGRIF_DEMO tracer.stat restartability FAILED : 15574 (results are different after 12 time steps)
WED025 run.stat restartability passed : 15574
ISOMIP+ run.stat restartability passed : 15574
OVERFLOW run.stat restartability passed : 15574
LOCK_EXCHANGE run.stat restartability passed : 15574
VORTEX run.stat restartability FAILED : 15574 (results are different after 121 time steps)
ICE_AGRIF ocean.output MISSING : 15574
ICE_AGRIF incomplete test
SWG run.stat restartability passed : 15574
!----repro----!
GYRE_PISCES run.stat reproducibility passed : 15574
GYRE_PISCES tracer.stat reproducibility passed : 15574
ORCA2_ICE_PISCES run.stat reproducibility passed : 15574
ORCA2_ICE_PISCES tracer.stat reproducibility passed : 15574
ORCA2_OFF_PISCES tracer.stat reproducibility passed : 15574
AMM12 ocean.output MISSING : 15574
AMM12 incomplete test
ORCA2_SAS_ICE run.stat reproducibility passed : 15574
ORCA2_ICE_OBS run.stat reproducibility passed : 15574
AGRIF_DEMO run.stat reproducibility passed : 15574
AGRIF_DEMO tracer.stat reproducibility passed : 15574
WED025 run.stat reproducibility passed : 15574
ISOMIP+ run.stat reproducibility passed : 15574
VORTEX run.stat reproducibility passed : 15574
ICE_AGRIF ocean.output MISSING : 15574
ICE_AGRIF incomplete test
SWG run.stat reproducibility passed : 15574
!----agrif check----!
ORCA2 AGRIF vs ORCA2 NOAGRIF run.stat unchanged - passed : 15574 15574
comment:41 Changed 3 years ago by techene
In 15581:
comment:42 Changed 3 years ago by techene
In 15586:
comment:43 Changed 3 years ago by techene
./sette.sh -At
Current code is : NEMO/branches/2021/dev_r14318_RK3_stage1 @ r15576 ( last change @ r15574 )
SETTE validation report generated for :
NEMO/branches/2021/dev_r14318_RK3_stage1 @ r15574+ (last changed revision)
on X64_IRENE arch file
!!---------------1st pass------------------!!
!----restart----!
GYRE_PISCES run.stat restartability passed : 15574+
GYRE_PISCES tracer.stat restartability passed : 15574+
ORCA2_ICE_PISCES run.stat restartability passed : 15574+
ORCA2_ICE_PISCES tracer.stat restartability passed : 15574+
ORCA2_OFF_PISCES tracer.stat restartability passed : 15574+
AMM12 run.stat restartability passed : 15574+
ORCA2_SAS_ICE run.stat restartability passed : 15574+
AGRIF_DEMO run.stat restartability passed : 15574+
AGRIF_DEMO tracer.stat restartability passed : 15574+
WED025 run.stat restartability passed : 15574+
ISOMIP+ run.stat restartability passed : 15574+
OVERFLOW run.stat restartability passed : 15574+
LOCK_EXCHANGE run.stat restartability passed : 15574+
VORTEX run.stat restartability passed : 15574+
ICE_AGRIF run.stat restartability passed : 15574+
SWG run.stat restartability passed : 15574+
!----repro----!
GYRE_PISCES run.stat reproducibility passed : 15574+
GYRE_PISCES tracer.stat reproducibility passed : 15574+
ORCA2_ICE_PISCES run.stat reproducibility passed : 15574+
ORCA2_ICE_PISCES tracer.stat reproducibility passed : 15574+
ORCA2_OFF_PISCES tracer.stat reproducibility passed : 15574+
AMM12 run.stat reproducibility passed : 15574+
ORCA2_SAS_ICE run.stat reproducibility passed : 15574+
ORCA2_ICE_OBS run.stat reproducibility passed : 15574+
AGRIF_DEMO run.stat reproducibility passed : 15574+
AGRIF_DEMO tracer.stat reproducibility passed : 15574+
WED025 run.stat reproducibility passed : 15574+
ISOMIP+ run.stat reproducibility passed : 15574+
VORTEX run.stat reproducibility passed : 15574+
ICE_AGRIF run.stat reproducibility passed : 15574+
SWG run.stat reproducibility passed : 15574+
comment:44 Changed 3 years ago by techene
In 15592:
comment:45 Changed 3 years ago by techene
In 15620:
comment:46 Changed 3 years ago by techene
In 15621:
comment:47 Changed 3 years ago by techene
In 15622:
comment:48 Changed 2 years ago by techene
In 15623:
comment:49 Changed 2 years ago by techene
In 15625:
comment:50 Changed 2 years ago by techene
In 15626:
comment:51 Changed 2 years ago by techene
In 15627:
comment:52 Changed 2 years ago by techene
In 15628:
comment:53 Changed 2 years ago by techene
In 15630:
comment:54 Changed 2 years ago by nemo
- Milestone changed from IMMERSE 2021 to Unscheduled
In 14353: