ChowPatelIlu.hpp
Go to the documentation of this file.
29// Pass -DUSE_CHOW_PATEL_ILU=1 to cmake to define CHOW_PATEL and use the iterative ILU decomposition
31// Pass -DUSE_CHOW_PATEL_ILU_GPU_PARALLEL=1 to use more parallelisation in the GPU kernel, see ChowPatelIlu.cpp
35// FINE-GRAINED PARALLEL INCOMPLETE LU FACTORIZATION, E. Chow and A. Patel, SIAM 2015, https://doi.org/10.1137/140968896
40// ChowPatelIlu could also operate on a full matrix LU when L and U are merged, but it is generally better to keep them split
51// This class implements a blocked version on GPU of the Fine-Grained Parallel ILU (FGPILU) by Chow and Patel 2015:
52// FINE-GRAINED PARALLEL INCOMPLETE LU FACTORIZATION, E. Chow and A. Patel, SIAM 2015, https://doi.org/10.1137/140968896
54// decomposition() allocates the cl::Buffers on the first call, these are C++ objects that deallocate automatically
Definition: blackoilboundaryratevector.hh:37 |