Closed zhangchonglin closed 2 years ago
Since within the function rebuild
, rebuild
time cost is always counted for every function call, time cost for rebuild
should be fine.
https://github.com/SCOREC/pumi-pic/blob/a710faa7f0982ce2d8656585efaec9d1c5fec29d/particle_structs/src/scs/SCS_rebuild.h#L300
Looking at the
migrate
function withinPUMIPic
:https://github.com/SCOREC/pumi-pic/blob/fea3a38afd490a69dc8a597cee05ab31c00505b8/particle_structs/src/scs/SCS_migrate.h#L211-L222
the time cost of
rebuild
particle structure withinmigrate
function is deducted from the totalmigration
time cost. It's fine if this is the PUMIPic definition ofparticle migration
time cost. But therebuild
time cost should be added to the overall particle structurerebuild
time cost in PUMIPic, instead of not counted. This seems to be the case forSCS
,CSR
, andCabanaM
particlemigrate
functions.Refer to https://github.com/SCOREC/xgcm/issues/154 for further details.