| ... | ... | @@ -18,7 +18,7 @@ The fist approach try to create a model for computation and comunication and try |
|
|
|
|
|
|
|
The second just divide the domain as the user specified and assign 1 sub-domain to each processor
|
|
|
|
|
|
|
|
## Model decomposition
|
|
|
|
### Model decomposition
|
|
|
|
|
|
|
|
In a distributed-memory setting, where data are scattered across
|
|
|
|
processors, two factors are important: equal division of work across
|
| ... | ... | @@ -37,13 +37,14 @@ the decomposition can be post-process further more to be for optimal, based on f |
|
|
|
|
|
|
|
* [CartDecomposition](http://ppmcore.mpi-cbg.de/doxygen/openfpm_pdata/classCartDecomposition.html)
|
|
|
|
|
|
|
|
## Fixed decomposition
|
|
|
|
### Fixed decomposition
|
|
|
|
|
|
|
|
The decomposition is chosen by the user and kept fixed.
|
|
|
|
|
|
|
|
* [CartDecomposition](http://ppmcore.mpi-cbg.de/doxygen/openfpm_pdata/classCartDecomposition.html)
|
|
|
|
|
|
|
|
## Data Driven decomposition
|
|
|
|
### Data Driven decomposition
|
|
|
|
|
|
|
|
Data is divided across processors, It is the default decomposition when a decomposition is not available yet, it basically divide the data consistently across processors regardless of geometrical meaning or communication
|
|
|
|
|
|
|
|
## Ghost |
|
|
\ No newline at end of file |
