2.8.3

What's Changed

• Fix/meshpart hierarchy tests by @suvarchal in #1120
• Improve scaling via Forcing: per-rank box scatter as an alternative to the full-grid broadcast by @suvarchal in #1119
• keep press_air under ice shelves by @chrisdane in #1121
• KPP/LF17: fix NaN in unresolved shear causing bottom-reaching OBL in the tropics by @patrickscholz in #1122
• Move step log and blow-up check into a post-step monitor by @suvarchal in #1107
• Conservation budgets: volume, heat and salt against the surface fluxes by @suvarchal in #1117
• Allocate helem and bottom_elem_thickness with the element halo size by @suvarchal in #1123
• clean up includes by @suvarchal in #1124
• Replace associate_*.h pointer aliases in the OpenACC files by @suvarchal in #1125
• Convert cpl_driver and remove the associate_*.h pointer alias headers by @JanStreffing in #1128
• fix: TOY mixing scheme collides with cvmix_TIDAL's internal id by @patrickscholz in #1129
• Send surface currents to the atmosphere in geographical coordinates by @JanStreffing in #1133
• Correct output unit strings and descriptions of unused switches by @JanStreffing in #1131
• Add missing initial values, namelist read checks and guards by @JanStreffing in #1132
• Backscatter: average V_work from V_work in uke_update by @JanStreffing in #1134
• Allocate fer_tapfac also for Redi without GM by @JanStreffing in #1135
• GM: divide the mixed-layer mean of bvfreq by the number of summed levels by @JanStreffing in #1137
• CMOR pbo: do not count the surface elevation twice with zlevel and zstar by @JanStreffing in #1138
• DVD: scale the surface flux term at the first wet level by @JanStreffing in #1139
• Ice diagnostics: vol_ice and vol_snow without the extra concentration factor by @JanStreffing in #1140
• Output: register the tracer flux streams also when TRGRD_XYZ is requested by @JanStreffing in #1141
• Tides: start the tidal clock at the model day and time by @JanStreffing in #1142
• OpenACC: make the GPU kernels build with nvfortran by @suvarchal in #1143
• OpenACC: correct and reproducible GPU results (sea ice, tracer update) by @suvarchal in #1144
• Split-explicit: stop at start-up with cavities or zlevel by @JanStreffing in #1146
• CMOR sea ice: split the hemispheres by geographical latitude by @JanStreffing in #1148
• Transient tracers: compute the wind speed from the magnitude of the stress by @JanStreffing in #1179
• Redi: fix the surface coefficient scaling and its reset under cavities by @JanStreffing in #1136
• Reference density: use the quadratic pressure term of the bulk modulus by @JanStreffing in #1177
• Blow-up check: make the sea surface elevation limit a namelist parameter by @JanStreffing in #1178
• Open-water albedo: compute it per vertex and keep the shared albw unchanged by @JanStreffing in #1180
• Restart: read the raw restart only if raw restarts are switched on by @JanStreffing in #1181
• Sea ice: stop if new_iclasses is used with a class count other than 15 by @JanStreffing in #1182
• DVD: do not modify the model velocities; implicit terms only with i_vert_diff by @JanStreffing in #1183
• Icepack: save the previous ice volume from m_ice, not from a_ice by @JanStreffing in #1184
• Backscatter: start the edge and element loops at the first wet level by @JanStreffing in #1185
• dMOC: bin the surface fluxes at the first wet level under ice shelves by @JanStreffing in #1186
• XIOS template: let the element files use the collective default by @JanStreffing in #1175
• CMOR: include the layer thickness change in the heat and salt tendencies by @JanStreffing in #1187
• Heat content: include the layer that contains the target depth in hc300m and hc700m by @JanStreffing in #1188
• IDEMIX2: include the x-gradient of f in the compartment refraction by @JanStreffing in #1189
• Single precision conservation by @suvarchal in #1206
• Single precision: layer thickness in real64 by @suvarchal in #1207
• par_ex always aborts; par_finalize for the normal end of a run by @JanStreffing in #1190
• Fix shortwave penetration depth relative to the ALE sea surface by @YuchenSun33 in #1116
• Increase GPU: OpenACC coverage after #1144 by @suvarchal in #1194
• ice_thermo_cpl: take the lead closing minimum from h0/h0_s by @sebastianbeyer in #1210
• nvfortran: restart and output of derived-type sections, tracer list namelist by @suvarchal in #1195
• Tracers on the GPU: AB history, Redi and implicit vertical diffusion, tracer gradients; packed EVP halo exchange by @suvarchal in #1196
• SSH conjugate-gradient solver and momentum (rhs, backscatter filter, implicit viscosity) on the GPU by @suvarchal in #1197
• Density, pressure gradient force, neutral slopes and vertical velocity on the GPU by @suvarchal in #1198
• KPP smoothing, tracer advection gathers, GM/Redi coefficients and bolus velocity, SSH rhs and hbar on the GPU by @suvarchal in #1199
• SSH stiffness matrix, KPP vertical mixing and nonlocal terms on the GPU; fewer tracer-step uploads by @suvarchal in #1200
• Monitor, layer thickness, neutral-slope downloads, NCAR bulk formulae and mo_convect on the GPU by @suvarchal in #1201
• Coalesced column kernels, one thread per CG row, sea-ice thermodynamics on the GPU by @suvarchal in #1202
• Velocity update, nodal velocities and output means on the GPU; packed 3D halo exchange by @suvarchal in #1203
• OpenACC: one build switch (ENABLE_OPENACC) checked against the compiler; NVHPC flags from the CMake options, for the library too, with -Mbackslash by @suvarchal in #1204
• Initialise ltra_diag in tracer_init and set it in the shipped namelists by @JanStreffing in #1211

New Contributors

• @YuchenSun33 made their first contribution in #1116

Full Changelog: 2.8.2...2.8.3