As of CY50 (and in specific branches on older cycles), PGD now should be able to run in parallel (MPI). It would be interesting to set Davai PGD tests parallel to run faster.
It requires:
to compile with macro -DSFX_MPI
to run with namelist key LSFX_MPI=.T. (unless this is the default in PGD and PREP ? to be checked)
to switch attribute engine="blind" to engine="parallel" in the AlgoComponent of the test script and set up a number of MPI tasks (?)
Another proposal is to set NHALO_PREP=20 in PREP to speed it up (cf. comment in #26)
As of CY50 (and in specific branches on older cycles), PGD now should be able to run in parallel (MPI). It would be interesting to set Davai PGD tests parallel to run faster.
It requires:
-DSFX_MPILSFX_MPI=.T.(unless this is the default in PGD and PREP ? to be checked)engine="blind"toengine="parallel"in the AlgoComponent of the test script and set up a number of MPI tasks (?)Another proposal is to set
NHALO_PREP=20in PREP to speed it up (cf. comment in #26)