Changeset 14200 for NEMO/branches/2020/dev_r14116_HPC-04_mcastril_Mixed_Precision_implementation_final/src/OCE/TRA/traldf_lap_blp.F90

Timestamp:

2020-12-17T15:36:44+01:00 (4 years ago)

Author:

mcastril

Message:

Merging r14117 through r14199 into dev_r14116_HPC-04_mcastril_Mixed_Precision_implementation_final

File:

• NEMO/branches/2020/dev_r14116_HPC-04_mcastril_Mixed_Precision_implementation_final/src/OCE/TRA/traldf_lap_blp.F90 (modified) (2 diffs)

NEMO/branches/2020/dev_r14116_HPC-04_mcastril_Mixed_Precision_implementation_final/src/OCE/TRA/traldf_lap_blp.F90

@@ -158 +158 @@
          ENDIF
          !
-         ! NOTE: [tiling-comms-merge] Bounds changed to avoid repeating this calculation for overlapping rows when using tiling
          DO_3D( isj, iej, isi, iei, 1, jpkm1 )            !== Second derivative (divergence) added to the general tracer trends  ==!
             pt_rhs(ji,jj,jk,jn) = pt_rhs(ji,jj,jk,jn) + (  ztu(ji,jj,jk) - ztu(ji-1,jj,jk)     &
@@ -236 +235 @@
       END SELECT
       !
-      ! NOTE: [tiling-comms-merge] Needed for both nn_hls as tra_ldf_iso and tra_ldf_triad have not yet been adjusted to work with nn_hls = 2. In the zps case the lbc_lnk in zps_hde handles this, but in the zco case zlap always needs this lbc_lnk. I did try adjusting the bounds in tra_ldf_iso and tra_ldf_triad so this lbc_lnk was only needed for nn_hls = 1, but this was not correct and I did not have time to figure out why
       CALL lbc_lnk( 'traldf_lap_blp', zlap(:,:,:,:) , 'T', 1.0_wp )     ! Lateral boundary conditions (unchanged sign)
       !                                               ! Partial top/bottom cell: GRADh( zlap )