Auto select memory-efficient algorithms for HMG #16210
Labels
C: Framework
P: normal
A defect affecting operation with a low possibility of significantly affects.
T: task
An enhancement to the software.
Reason
HMG
is a high-performance (hybrid)MG
preconditioner that uses HYPRE to coarsen matrix and PETSc PCs as the smoothers. It is able to reuse interpolation and restriction matrices of one moose variable for all other nonlinear moose variables. Use interpolation and restriction matrices to form coarse matrices requires memory-efficient algorithms. The default algorithms in PETSc generate unexpected memory spikes. We need to use new algorithms "allatonce
".Design
Automatically apply PtAP algorithms when users use HMG.
Impact
Reduce memory usage for large scale problems
The text was updated successfully, but these errors were encountered: