Review DynEarthSol #114 for doc updates: Regular update 2026-10: quieter MMG remesh log, device-side averaged output, two OpenACC races, 2vtk marker sets #102

GeoFLAC/DynEarthSol#114 merged — Regular update 2026-10: quieter MMG remesh log, device-side averaged output, two OpenACC races, 2vtk marker sets

Why

Five small changes collected since #95 and #110. An MMG remesh filled the screen with 10-18 bookkeeping lines per remesh; averaged output (the default) touched four fields on the host every step, which under managed memory moved their pages off the GPU every step; two async OpenACC kernels could still be running when the host read or freed what they used; 2vtk.py never converted the hydrous marker set of an hdf5=1 run; and an info_display_step_interval that is not a multiple of quality_check_step_interval was rounded up and then rejected anyway. All five leave model output bit-identical (see Numerical effect).

No related issue.

What

Files changed

• 2vtk.py
• CHANGELOG.md
• Dynearthsol.py
• dynearthsol.cxx
• examples/core-complex-mmg.cfg
• fields.cxx
• geometry.cxx
• input.cxx
• markerset.cxx
• mesh.cxx
• mmg_utils.cxx
• output.cxx
• remeshing.cxx
• remeshing.hpp

Next step

Ask Claude Code to run the doc-sync procedure from CLAUDE.md against this PR.