Vectorized Stoke Solver

[andrewdnolan]

Implementing the Vectorized (and threaded) Stokes Solver (Stokes-Vec) would result in significant simulation speed ups. There have be numerous reasons barring us from its implementation, some resolved some still unresolved.

Resolved Reasons:

• .result file parsing:
  • An early attempt use Stokes-Vec was abandoned because it wrote it's solution vectors to .result files differently, than the legacy Stokes solver. elmer2nc was not built out to handle these differences, so Stokes-Vec was abandoned.
  • With the adoption of NetcdfUGRIDOutputSolver Solver, this no longer an obstacle. NetCDF files with solution vectors are directly written by the .sif file so don't need to parse .result files anymore.

Unresolved Reasons:

• Structured Mesh Mapper:
  • It appears as though the Stokes-Vec solver does not work well with the Structured Mesh Mapper.