Particle-grid migrate #395
Conversation
streeve
force-pushed
the
pg_dist
branch
2 times, most recently
from
83cec77 to
b8e0c9d
Compare
streeve
force-pushed
the
pg_dist
branch
2 times, most recently
from
3019a5c to
67afc76
Compare
Codecov Report
@@ Coverage Diff @@
## master #395 +/- ##
======================================
Coverage 93.0% 93.0%
======================================
Files 36 37 +1
Lines 4194 4269 +75
======================================
+ Hits 3902 3973 +71
- Misses 292 296 +4
Continue to review full report at Codecov.
|
|
Ok @sslattery this is ready for review. I will keep working on the similar arbitrary dimension halo functionality and create a separate PR |
|
|
||
| for ( std::size_t d = 0; d < num_space_dim; ++d ) | ||
| { | ||
| local_low[d] = local_mesh.lowCorner( Cajita::Own(), d ); |
There was a problem hiding this comment.
We do this in a lot of places. No need to add in this PR but it would be nice to have functions for these that do not take a dimension argument and just return a Kokkos::Array for all dimensions.
| destinations( p ) = neighbor_ranks( neighbor_index ); | ||
|
|
||
| // Shift particles through periodic boundaries. | ||
| for ( std::size_t d = 0; d < num_space_dim; ++d ) |
There was a problem hiding this comment.
nice - good kernel fusion. I think we can also eventually find a way to fuse in the count kernel as well so we have only a single setup kernel.
|
@sslattery should be ready for final review |
This adds particle-grid migrate in Cajita, wrapping calls to
Cabana::Distributor,Cabana::migrateparticleGridMigrateuses implementation functions and callsmigrateThis supersedes #263 and is part of the very old #77