MatPtAP: use MatSeqAIJGetArray to access arrays, to be device safe

Fix long-standing bug in MatMPIAIJDiagonalScaleLocalSetUp()

This commit fixes a problem I discovered in October 2017 and then forgot about
after some unsuccessful initial attempts to find the proble

Fix long-standing bug in MatMPIAIJDiagonalScaleLocalSetUp()

This commit fixes a problem I discovered in October 2017 and then forgot about
after some unsuccessful initial attempts to find the problem. The MPIAIJ
implementation for MatDiagonalScaleLocalSetUp() was broken when the
local-to-global mappings in PETSc were refactored to always use
block indices; the MPIAIJ implementation of
MatDiagonalScaleLocalSetUp(), however, used element-wise, not block
indexing, and this was not refactored at the time, as it was overlooked
due to having no test case.
My commit finally fixes this; it also fixes several memory leaks in the
AIJ and BAIJ implementations of MatDiagonalScaleLocal() which were
overlooked, again, due to having no test cases.

show more ...

Kokkos: update to v5.0.0

CI: remove kokkos from linux-cuda-pkgs (in favor of linux-kokkos-cuda) as v5 breaks with many external pkgs,
CI: continue to use kokkos v4.7.01 for hip builds - as they curr

Kokkos: update to v5.0.0

CI: remove kokkos from linux-cuda-pkgs (in favor of linux-kokkos-cuda) as v5 breaks with many external pkgs,
CI: continue to use kokkos v4.7.01 for hip builds - as they currently use rocm-5.4.3

Co-authored-by: Satish Balay <balay@mcs.anl.gov>

show more ...

Merge branch 'release'

superlu-dist: superlu_acc_offload is available in v8.0.0+

Merge remote-tracking branch 'origin/release'

Fix builds against CUDA 13.1

- Fix usage of deprecated thrust:: APIs in CCCL 3.1
- Explicitly include "thrust/copy.h" to fix builds w/ CUDA 13.1
- configure: use SuperLU_DIST 9.2.1 for CUDA 13.1
- C

Fix builds against CUDA 13.1

- Fix usage of deprecated thrust:: APIs in CCCL 3.1
- Explicitly include "thrust/copy.h" to fix builds w/ CUDA 13.1
- configure: use SuperLU_DIST 9.2.1 for CUDA 13.1
- CI: update linux-cuda-latest to test with CUDA 13.1

show more ...

Merge branch 'tcaruso/mumps_ICNTL_49' into 'main'

Add support for MUMPS ICNTL(49) in PETSc

See merge request petsc/petsc!8886

MATSOLVERMUMPS: add support for ICNTL(49)

Funded-by: France 2030 NumPEx Exa-MA (ANR-22-EXNU-0002) project

MATSOLVERMUMPS: fix compilation with OpenMP and no mixed precision

Rework MAT_SYMMETRIC and MAT_HERMITIAN options

Merge branch 'release'

Add option to disable SuperLU_DIST GPU offloading

MatAssemblyEnd(): fix missing space in -info

Mat: set appropriate NO_OFF_PROC_ENTRIES to bypass MPI_Allreduce()

Fix typos

Merge branch 'jolivet/fix-nonscalable-matnorm' into 'main'

MatNorm: fix some non-scalable implementations for NORM_1

See merge request petsc/petsc!8862

Merge branch 'jose/warning-hip' into 'main'

Fix HIP compile warning

See merge request petsc/petsc!8861

VecSetValues() instead

MatNorm: fix some non-scalable implementations for NORM_1

Reported-by: Vincent Robert <vincent.robert@cea.fr>

Fix HIP compile warning

src/mat/impls/aij/seq/seqhipsparse/aijhipsparse.hip.cpp:1401:15: error: unused variable 'i' [-Werror,-Wunused-variable]
1401 | PetscInt i;
| ^
src/m

Fix HIP compile warning

src/mat/impls/aij/seq/seqhipsparse/aijhipsparse.hip.cpp:1401:15: error: unused variable 'i' [-Werror,-Wunused-variable]
1401 | PetscInt i;
| ^
src/mat/impls/aij/seq/seqhipsparse/aijhipsparse.hip.cpp:1647:15: error: unused variable 'i' [-Werror,-Wunused-variable]
1647 | PetscInt i;
| ^

show more ...

Housekeeping

Merge branch 'barry/2025-10-18/optimize-aij-ops' into 'main'

Refactor handling of diagonal marking in *AIJ and SELL matrices

See merge request petsc/petsc!8797

Optimizations on Seq_[SB]AIJ operations that use MatGetDiagonalMarkers.

- Replace MatMissingDiagonal() and MatMarkDiagonals_SeqXXX() with MatGetDiagonalMarkers_SeqXXX()

- Mat_SeqXXX->diag is not automatically created during MatAssemblyEnd() saving memory and time

- Replace MatMissingDiagonal() and MatMarkDiagonals_SeqXXX() with MatGetDiagonalMarkers_SeqXXX()

- Mat_SeqXXX->diag is not automatically created during MatAssemblyEnd() saving memory and time

- Accessing Mat_SeqXXX->diag now requires the use of MatGetDiagonalMarkers_SeqXXX() except when the current values are known to be correct; for example during numerical factorizations and solves

- Mat_SeqXXX->diag is now never shared among matrices; hence the free_diag flag is gone. That was always a risky proposition since any of the owning matrices could chang the values thus making them incorrect for other owners.

show more ...