Yohann Dudouit
caa79bf801
feedback from @pazner
2022-08-31 17:17:18 -07:00
Yohann Dudouit
a49b8a5a63
Merge branch 'master' into yohann/sort+test
2022-08-25 13:35:30 -07:00
Yohann Dudouit
96f8aa4a89
Fix sorting of sparse matrices with hip.
2022-08-25 10:07:03 -07:00
Yohann Dudouit
d7b5d07cce
Create temporary cusparseMatDescr_t for sorting.
2022-08-23 17:37:46 -07:00
Yohann Dudouit
8029a5ace0
Use Arrays instead of Memory objects.
2022-08-23 17:37:14 -07:00
Yohann Dudouit
78ac894623
Use class methods to read and write.
2022-08-23 17:36:44 -07:00
Yohann Dudouit
6cd171ae05
Use Capacity instead of Size.
2022-08-23 17:33:30 -07:00
Yohann Dudouit
9f035181c1
Update SortColumnIndices to be device aware.
2022-08-23 15:27:28 -07:00
Prathamesh Sirmalla @lab_pc
1f3d8a051c
Minor change in output message in linalg/sparsemat.cpp
2022-08-19 15:38:25 -05:00
Yohann Dudouit
e19ef66f80
Avoid shadowing in SparseMatrix.
2022-06-13 11:15:55 -07:00
Yohann Dudouit
45d5bd70e1
Move ElimnateBC in SparseMatrix.
2022-06-13 10:57:56 -07:00
Tzanio Kolev
335ceb9a94
Merge branch 'master' into lor-batched
...
Conflicts:
fem/CMakeLists.txt
linalg/sparsemat.cpp
2022-05-08 14:17:49 -07:00
Will Pazner
a9f560ae72
Merge remote-tracking branch 'origin/master' into yohann/add_CUDA_tests
2022-04-20 10:12:22 -07:00
Yohann Dudouit
a621595c8c
Revert NumNonZeroElems() implementation.
2022-04-15 11:35:32 -07:00
Yohann Dudouit
1063ddc3d0
Delete alias to try to fix bug.
2022-03-23 12:56:20 -07:00
Yohann Dudouit
60505fae80
Optimize NumNonZeroElems when the SparseMatrix is on device.
2022-03-23 12:32:16 -07:00
Yohann Dudouit
78610bc996
Replace HostReadData with Finalized.
2022-03-23 12:18:45 -07:00
Yohann Dudouit
785b7cf0b1
Replace A != NULL with HostReadData() != NULL.
...
NOTE: This is very innefficient.
2022-03-23 11:50:12 -07:00
Yohann Dudouit
bcfe4c4258
Make NumNonZeroElems device compatible.
...
- Copy `I` on the host... not optimal at all.
2022-03-23 11:45:25 -07:00
Yohann Dudouit
c135722383
Make Transpose device compatible.
...
- The transpose is still built on host.
2022-03-23 11:41:21 -07:00
Will Pazner
b10a3486f7
Merge branch 'fa-mem-leak' into lor-batched-gpu-no-mesh
2022-03-21 11:29:24 -07:00
Will Pazner
52206d3416
Add SparseMatrix::OverrideSize
...
This is an "expert function" to change the height and width of a
SparseMatrix after it has been constructed.
2022-03-21 10:48:52 -07:00
Will Pazner
947422c8c3
SparseMatrix: check for Finalized rather than A != NULL
2022-03-02 10:01:02 -08:00
Will Pazner
1fb26145b1
Merge remote-tracking branch 'origin/master' into lor-batched-gpu-no-mesh
...
# Conflicts:
# general/cuda.hpp
# tests/unit/CMakeLists.txt
2022-03-01 17:54:15 -08:00
Tzanio
6d2530775b
Extended copyright to 2022
2022-02-25 10:24:50 -08:00
Tzanio Kolev
935d3f6df0
Merge pull request #2737 from mfem/shadow-linalg-dev
...
Fixing shadow variables in linalg classes [shadow-linalg-dev]
2022-02-24 08:50:38 -08:00
Stowell, Mark L
53e4896385
Merge remote-tracking branch 'origin/master' into shadow-linalg-dev
2022-02-18 09:16:27 -08:00
Will Pazner
14b48da4d2
Merge remote-tracking branch 'origin/master' into shadow-out-dev
2022-02-18 08:20:52 -08:00
Stowell, Mark L
e5558895dc
Suggestions made by @YohannDudouit
2022-02-17 16:02:57 -08:00
Stowell, Mark L
0ee82397de
Following suggestions from @pazner
2022-02-17 15:02:18 -08:00
Will Pazner
b8a3e423fd
Update condition in SparseMatrix::MultTranspose to match EnsureMultTranspose
2022-02-10 15:09:33 -08:00
Will Pazner
5bd76ca866
Add SparseMatrix::EnsureMultTranspose
2022-02-10 11:19:32 -08:00
Veselin Dobrev
9f7ab4e642
Merge branch 'master' into hipsparse
...
Resolved conflicts:
CHANGELOG
CMakeLists.txt
config/cmake/modules/MfemCmakeUtilities.cmake
Additional updated files:
config/defaults.mk
2022-02-09 11:53:54 -08:00
Veselin Dobrev
564d7aa1ca
In sparsemat.cpp, add MFEM_ prefix to the CUDA-or-HIP macros.
2022-02-05 15:47:57 -08:00
Tom Stitt
e33c78fda1
fix potentially missed cusparse/hipsparse cleanup - thanks Veselin
2022-01-27 16:41:36 -08:00
Tom Stitt
baffb28a4b
move substitution macros to sparsemat.cpp
...
define MFEM_USE_CUDA_OR_HIP in cuda.hpp and hip.hpp
readd and deprecate CleanCUSparse and UseCUSparse
2022-01-27 09:40:47 -08:00
Tom Stitt
aa5b1101b4
doxygen ignore macro-assisted typename
2022-01-25 09:20:20 -08:00
Tom Stitt
3b9000c617
add hipsparse support
2022-01-24 09:54:52 -08:00
Arturo Vargas
73916e5140
bugfix for sparse mult transpose
2022-01-20 15:31:45 -08:00
Stowell, Mark L
63381b366a
Renaming ostream variables from out to os
2022-01-18 15:06:46 -08:00
Stowell, Mark L
61215468f3
Fixing shadow variables in linalg classes
2021-12-23 16:25:02 -08:00
Stowell, Mark L
f7c9bdab17
Renaming "out" and "err" local variables to avoid globals
2021-12-22 15:19:20 -08:00
Veselin Dobrev
a58567695e
Fix an error when building with HIP which may also be causing
...
problems with CUDA.
In OperatorJacobiSmoother, use absolute value of diagonal, if
requested, during setup instead of during action.
A few tweaks in SparseMatrix::Jacobi and SparseMatrix::DiagScale.
For the GitLab CI on Lassen, disable ATS.
2021-12-14 22:37:53 -08:00
camierjs
fdb5d7d627
Parallel debug device
2021-11-30 15:53:26 -08:00
Vladimir Z Tomov
2d10dd0abe
Options to use abs values of the diagonal in OperatorJacobi, DSmoother.
2021-11-26 15:49:40 -08:00
Tzanio
0cc104331c
Merge branch 'master' into yohann/mem/min-default-constr
2021-11-10 12:52:15 -08:00
Will Pazner
233bee6188
Add comment to PrintMatlab
2021-09-28 14:17:59 -07:00
Will Pazner
9e67380171
Improvements to PrintMatlab
...
Make Operator::PrintMatlab virtual, output zero entry in last row and column to
preserve matrix size.
2021-09-28 12:01:04 -07:00
Tzanio Kolev
2ed61b9888
Merge pull request #2446 from mfem/jacobf/2021-08-02/feature/jacobi23-on-device
...
Feature: Jacobi2/3 device
2021-09-09 20:29:29 -07:00
Jacob Faibussowitsch
fa845400b1
style
2021-08-20 15:20:54 -05:00
Yohann Dudouit