| 8cde19ba | 06-Sep-2018 |
Richard Tran Mills <rmills@rmills.org> |
Fixed style errors identified by Barry, plus a few others identified by me. |
| d615724a | 05-Sep-2018 |
Richard Tran Mills <rmills@rmills.org> |
Removed duplicate definition of MAT_Convert. |
| 0e113e69 | 05-Sep-2018 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'stefano_zampini/feature-mataij-to-matis-rebased-final' |
| 2d4298ae | 09-Aug-2018 |
Junchao Zhang <jczhang@mcs.anl.gov> |
Style change: rename comm_mumps to mumps_comm, size to petsc_size |
| a1dfcbd9 | 09-Aug-2018 |
Junchao Zhang <jczhang@mcs.anl.gov> |
Improve uses of mumps->id.lsol_loc
1) mumps->id.INFO(23) was set by MUMPS in the numeric factorization phase and we stored in mumps->id.lsol_loc. We can reuse mumps->id.lsol_loc 2) Length of mump
Improve uses of mumps->id.lsol_loc
1) mumps->id.INFO(23) was set by MUMPS in the numeric factorization phase and we stored in mumps->id.lsol_loc. We can reuse mumps->id.lsol_loc 2) Length of mumps->id.isol_loc[] is lsol_loc (MUMPS-5.1.2 Manual, page 46), not nrhs*lsol_loc. We do not need to allocate excessive memory
show more ...
|
| 0f52d626 | 09-Aug-2018 |
Junchao Zhang <jczhang@mcs.anl.gov> |
Fix a typo |
| dcf7d9d1 | 09-Aug-2018 |
Junchao Zhang <jczhang@mcs.anl.gov> |
Removed trailing spaces |
| ca9cdca7 | 04-Sep-2018 |
Richard Tran Mills <rmills@rmills.org> |
Added MATMPIAIJSELL matrix type. |
| 55c1dcbe | 04-Sep-2018 |
Satish Balay <balay@mcs.anl.gov> |
Merge remote-tracking branch 'origin/pr1095/aselinger/faster_MatTranspose_MPIAIJ/master' |
| 8197ac24 | 04-Sep-2018 |
Richard Tran Mills <rmills@rmills.org> |
Improved the description of MatCreateSeqAIJSELL() in its manual page. |
| 597d2e73 | 04-Sep-2018 |
Richard Tran Mills <rmills@rmills.org> |
Added MatConvert logging inside the SEQAIJSELL "shadow" matrix construction. |
| af22a668 | 03-Jul-2018 |
Richard Tran Mills <rmills@rmills.org> |
Added MatSOR_SeqAIJSELL(). |
| 84920f4b | 03-Jul-2018 |
Richard Tran Mills <rmills@rmills.org> |
Added composed function MatPtAP_is_seqaijsell_C(). |
| 4da9d7bd | 02-Jul-2018 |
Richard Tran Mills <rmills@rmills.org> |
Add MultTranspose, MultAdd, MultTransposeAdd operations for AIJSELL. |
| 63663694 | 28-Jun-2018 |
Richard Tran Mills <rmills@rmills.org> |
Add "eager shadow" option for AIJSELL. |
| 9d769819 | 26-Jun-2018 |
Richard Tran Mills <rmills@rmills.org> |
MatMult_SeqAIJSELL() should call MatMult_SeqSELL() directly, rather than via MatMult(). |
| a2a1850b | 21-Jun-2018 |
Richard Tran Mills <rmills@rmills.org> |
Moved check to for whether rebuild is necessary into MatSeqAIJSELL_build_shadow(). |
| b0e5de86 | 21-Jun-2018 |
Richard Tran Mills <rmills@rmills.org> |
Implemented MatMult() for MATSEQAIJSELL. |
| 4dfdc2d9 | 20-Jun-2018 |
Richard Tran Mills <rmills@rmills.org> |
Added skeleton of sequential AIJSELL matrix class.
AIJSELL is a subclass of AIJ that uses AIJ for most operations, but keeps a "shadow" copy of the matrix in SELL format, which is used for certain o
Added skeleton of sequential AIJSELL matrix class.
AIJSELL is a subclass of AIJ that uses AIJ for most operations, but keeps a "shadow" copy of the matrix in SELL format, which is used for certain operations for which SELL format is preferable. This is currently a skeleton that doesn't do anything but call AIJ routines, but is ready for addition of optimized operations using SELL.
show more ...
|
| 59f888c2 | 04-Sep-2018 |
Stefano Zampini <stefano.zampini@gmail.com> |
MAT_MKLPARDISO: match check for data, not for Schur Mat |
| f79cb1a0 | 03-Sep-2018 |
andi selinger <andreasselinger@gmail.com> |
added conversion from void* to Mat_MPIAIJ* |
| a8661f62 | 31-Aug-2018 |
andi selinger <andreasselinger@gmail.com> |
use MatTranspose_SeqAIJ within MatTranspose_MPIAIJ |
| fefec95c | 31-Aug-2018 |
Barry Smith <bsmith@mcs.anl.gov> |
Merge branch 'barry/feature-PCVPBJACOBI' |
| e83a5fba | 31-Aug-2018 |
Barry Smith <bsmith@mcs.anl.gov> |
Merge branch 'maint' |
| a17e4e36 | 31-Aug-2018 |
Barry Smith <bsmith@mcs.anl.gov> |
Remove incorrect error check that only MATSEQDENSE is supported since MATMPIDENSE also supported
Note that the call to MatDenseGetArray(Read) on the X and B matrices will check that they are dense f
Remove incorrect error check that only MATSEQDENSE is supported since MATMPIDENSE also supported
Note that the call to MatDenseGetArray(Read) on the X and B matrices will check that they are dense format and thus can work in this routine
Commit-type: bug-fix Reported-by: Marius Buerkle <mbuerkle@web.de>
show more ...
|