Commit Graph

746 Commits

Author SHA1 Message Date
Arne Morten Kvarving
dd92a7446c BlackoilModelParametersEbos: rename to BlackoilModelParameters 2024-02-01 09:05:40 +01:00
Bård Skaflestad
859db850c0
Merge pull request #5111 from lisajulia/feature/deterministic-indicesSyncer
Construct the matrices in the AMG hierarchy with deterministic indices.
2024-01-29 13:37:54 +01:00
Lisa Julia Nebel
6dfa9b1b76 Add comment on trailing return type with decltype that is used for detecting the signature of the coarsen function of the Dune::Amg::IndicesCoarsener 2024-01-26 14:29:19 +01:00
Lisa Julia Nebel
74608147c0 Add comment on trailing return type with decltype that is used for detecting the existence of setUseFixedOrder member function 2024-01-26 14:29:15 +01:00
Lisa Julia Nebel
5b50a4801f Construct the matrices in the AMG hierarchy (created in the twolevelmethod) with deterministic indices if possible.
The function 'coarsen' for the coarser level is called with the respective flag if it is defined. Then the matrices in the AMG hierarchy are constructed with deterministic indices.
If it is not defined yet, it is called without the flag and the matrices in the AMG hierarchy are constructed with non-deterministic indices.
2024-01-26 13:51:41 +01:00
Atgeirr Flø Rasmussen
2626fbb84b
Merge pull request #5002 from multitalentloes/add_cudilu
Add cudilu
2024-01-25 15:43:39 +01:00
Tobias Meyer Andersen
4b0dd54f15 Add CUDA implementation of the DILU
preconditioner. Uses graph coloring to exploit
parallelism in upper and triangular solves when
computing a diagonal approximate inverse of a
sparse matrix. Supports blocksizes up to 3.
2024-01-25 14:26:38 +01:00
Lisa Julia Nebel
60b0a33bd4 Construct the matrices in the AMG hierarchy (created in the PreconditionerFactory) with deterministic indices if possible.
The function 'setUseFixedOrder' is called if it is defined and the matrices in the AMG hierarchy are constructed with deterministic indices.
If it is not defined yet, it is not called and the matrices in the AMG hierarchy are constructed with non-deterministic indices.
2024-01-25 11:32:09 +01:00
Atgeirr Flø Rasmussen
7c2f01996a Add cpr_trueimpesanalytic option for LinearSolver parameter. 2024-01-24 13:03:18 +01:00
Atgeirr Flø Rasmussen
ae838d0b84
Merge pull request #5103 from hnil/trueimpesanalytic
add analytic true impes
2024-01-18 22:51:57 +01:00
hnil
bec5a233eb removed more warings and a bug for strange index settings 2024-01-18 18:25:50 +01:00
hnil
cced62edc0 fixed warning 2024-01-18 18:18:42 +01:00
hnil
be27e2388d fixed name in cpp file 2024-01-18 11:40:59 +01:00
hnil
1900379343 changes to adjust to review comments 2024-01-18 10:37:48 +01:00
Atgeirr Flø Rasmussen
a52bcf51b1 Initialize the communication object in initialize(). 2024-01-17 12:39:52 +01:00
hnil
61329532a7 -- add true impes
-- moved code to hpp to avoid unnesseary logic
2024-01-12 15:33:09 +01:00
Arne Morten Kvarving
f2a2fc2938 add missing include 2023-11-22 13:18:56 +01:00
Tobias Meyer Andersen
ee95223b27 Use std::optional to only allocate reordered matrix when using multiple threads 2023-11-22 10:10:29 +01:00
Tobias Meyer Andersen
e22f4399a0 fix typo 2023-11-22 09:16:43 +01:00
Tobias Meyer Andersen
5f6c97ff3b add OpenMP parallelized version of DILU.
Implement graphcoloring to expose rows in level sets that that can be
executed in parallel during the sparse triangular solves.
Add copy of A matrix that is reordered to ensure continuous memory reads
when traversing the matrix in level set order.
TODO: add number of threads available as constructor argument in DILU
2023-11-21 15:41:53 +01:00
Tobias Meyer Andersen
e2a531dd65 replace loop with std function, remove print from test 2023-11-21 09:52:48 +01:00
Tobias Meyer Andersen
3cd1873771 add graph coloring 2023-11-20 16:57:56 +01:00
Tobias Meyer Andersen
54b7f9c44f GraphColoring.hpp: cosmetics 2023-11-20 13:06:40 +01:00
Tobias Meyer Andersen
9309f96d8f GraphColoring.hpp: reorder headers 2023-11-20 13:06:34 +01:00
Antonella Ritorto
166c31d5f8 Bugfix. Deprecated getCollectiveComminication warning 2023-10-31 08:40:46 +01:00
Razvan Nane
45d8750e93 Fix formatting and indentation 2023-10-26 09:45:48 +02:00
Razvan Nane
cd63cc83c2 Remove redundant cpu decomposition when using gpu 2023-10-26 06:40:06 +02:00
Bård Skaflestad
7a34c20120 Remove Leftover Debug Printout
This information already goes to the .DBG file so there's no need
to echo it to the main console as well.
2023-10-24 09:51:12 +02:00
jakobtorben
ab0ca76194 Add DILU preconditioner 2023-10-18 14:30:17 +02:00
Atgeirr Flø Rasmussen
8b480a0a49
Merge pull request #4852 from multitalentloes/cujac_preconditioner
Cujac preconditioner
2023-10-16 12:46:29 +02:00
Tobias Meyer Andersen
8b396fa3dc Update test properties of new tests using CUDA. Avoid comparison causing new warning 2023-10-16 11:31:00 +02:00
Tobias Meyer Andersen
533c5a1a3e Combine new diagonalMV kernels into one using
template arguements.
Use more consistent naming conventions.
Change printing function of CuVector from output
stream overload to toDebugString().
2023-10-16 09:53:02 +02:00
Markus Blatt
e9bb5d0539
Merge pull request #4582 from Tongdongq/rocsparse-separate-wells
RocsparseSolver separate wells
2023-10-13 15:33:03 +02:00
Tobias Meyer Andersen
c809819cbd Fuse kernels to optimize performance and update
the tests and documentation.
2023-10-13 10:31:17 +02:00
Tobias Meyer Andersen
fbeb006d45 Simplify and optimize the Jacobi preconditioner
apply. Fix .h and .cpp mistakes in cmake.
2023-10-13 10:31:17 +02:00
Tobias Meyer Andersen
1e4b0e97ee Add jacobi preconditioner that runs on the GPU.
Implement calls to cuBlas, cuSparse and implement necessary
CUDA kernels to perform a single iteration of the jacobi preconditioner.

Add tests that verify new kernels and the preconditioner in its totality.

The preconditioner is verified on 2x2 and 3x3 blocks, which as of now
are the only supported sizes. 1x1 are not supported because cuSparse
does not support it.
2023-10-13 10:31:17 +02:00
Arne Morten Kvarving
3e047ab9cd
Merge pull request #4927 from blattms/spelling-fix1
Fixed spelling: achived -> achieved
2023-10-13 08:25:07 +02:00
Markus Blatt
50815209a0 Fixed spelling: achive -> achieved 2023-10-12 22:01:12 +02:00
Kjetil Olsen Lye
628b69c83c
Casting to support {fmt} v10.
Due to issues with anonymous enums, newer versions of {fmt} (v10) do not seem to handle the implicit conversion from enum value to `int`. Hence we need a cast for `fmt::format`-calls involving `Matrix::block_matrix::rows` (or `cols`) to compile. This seems to only be relevant for one part of the code. 

For an isolated example, see https://github.com/kjetilly/fmt_fails_with_enum
2023-10-10 10:58:21 +02:00
Kjetil Olsen Lye
df0fdbbd0b
Removed call to deprecated cudaThreadExit
`cudaThreadExit` has been deprecated in favour of `cudaDeviceReset`, which is already being called.
2023-10-09 10:49:31 +02:00
Razvan Nane
dcc3bd70f0 Reinclude changes lost in adaptation to ISTLSolverEbosBda 2023-10-08 08:58:35 +02:00
Razvan Nane
e4abc12a05 Adapt rocsparse separate wells PR to changes made to ISTLSolverEbos 2023-10-08 08:58:35 +02:00
Razvan Nane
177a46366d Add <rocsparse> to accelerator-mode option in doc 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
b2fea287bc Print time to apply wells for verbosity>=3 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
abce3a897c Only compile HIP kernel with hipcc 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
ef6be5859e Restore old HAVE_CUDA value 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
684397129b Make sure rocsparse can get wellcontributions 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
cb20d041c3 Allow WellContributionsRocsparse to be used 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
50ccea0880 Allow rocsparseSolver to choose deviceID 2023-10-08 08:58:35 +02:00
Tong Dong Qiu
9bef10a018 Set hipStream for rocsparseWellcontributions 2023-10-08 08:58:35 +02:00