| 75480915 | 05-Aug-2019 |
Pierre Jolivet <pierre.jolivet@enseeiht.fr> |
Initial support for SeqSBAIJ matrices. |
| 0cd7f59a | 01-Aug-2019 |
Barry Smith <bsmith@mcs.anl.gov> |
Fix bug in MatMPISBAIJSetPreallocationCSR()
Add test case that detected bug
Fix over-preallocation that occured in MatMPISBAIJSetPreallocationCSR() and MatSeqSBAIJSetPreallocationCSR()
Added docum
Fix bug in MatMPISBAIJSetPreallocationCSR()
Add test case that detected bug
Fix over-preallocation that occured in MatMPISBAIJSetPreallocationCSR() and MatSeqSBAIJSetPreallocationCSR()
Added documentation MATMPI[B]AIJ and MATSEQ[B]AIJ noting their support of two features
Commit-type: optimization, bug-fix, testing, documentation, example
Reported-by: Pierre Jolivet <pierre.jolivet@enseeiht.fr>
show more ...
|
| 082a2362 | 03-Aug-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'karlrupp/fix-cuda-empty-procs' [PR #1938]
* karlrupp/fix-cuda-empty-procs:
CUDA: added guards for empty process triangular solves |
| 504af54e | 03-Aug-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'karlrupp/fix-cuda-vecset' [PR #1937]
* karlrupp/fix-cuda-vecset:
Remove unneeded collective VecSet from a MatMultTranspose_SeqAIJCUSPARSE |
| cf00fe3b | 02-Aug-2019 |
Karl Rupp <me@karlrupp.net> |
CUDA: added guards for empty process triangular solves
Cherry-pick of c9cf7f9
Thanks-to: Mark Adams <ma2325@columbia.edu> |
| 2b551a2f | 03-Jul-2019 |
Mark Adams <ma2325@columbia.edu> |
use non-collective VecSet |
| a8bd5306 | 09-Jul-2019 |
Mark Adams <ma2325@columbia.edu> |
fixed bug |
| eef58048 | 02-Aug-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'hannah/gpu-logging-WaitForGPU' [PR #1927]
* hannah/gpu-logging-WaitForGPU:
Adding WaitForGPU() to GPU time
This branch rearranges GPU timers so that calls to WaitForGPU() are counted t
Merge branch 'hannah/gpu-logging-WaitForGPU' [PR #1927]
* hannah/gpu-logging-WaitForGPU:
Adding WaitForGPU() to GPU time
This branch rearranges GPU timers so that calls to WaitForGPU() are counted towards time spend on the GPU.
show more ...
|
| 34bcad68 | 01-Aug-2019 |
Fande Kong <fdkong.jd@gmail.com> |
MPIPtAP: bug fixes |
| ae28c693 | 26-Jun-2019 |
Fande Kong <fdkong.jd@gmail.com> |
Print what algorithm is used for MatPtAP. |
| bc8e477a | 26-Jun-2019 |
Fande Kong <fdkong.jd@gmail.com> |
multicomponent version of allatonce and allatonce_merged work |
| 5c65b9ec | 24-Jun-2019 |
Fande Kong <fdkong.jd@gmail.com> |
A better way to extract remote rows of P (interpolations) |
| 661c2d29 | 31-Jul-2019 |
hannah_mairs <hannah.mairs@gmail.com> |
Adding WaitForGPU() to GPU time |
| 78af062c | 31-Jul-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'barry/2019-07-30/add-use-petscintcast-corrected' [PR #1924]
* barry/2019-07-30/add-use-petscintcast-corrected:
Add PetscIntCast() to check for integer overflow |
| c73702f5 | 31-Jul-2019 |
Barry Smith <bsmith@mcs.anl.gov> |
Add PetscIntCast() to check for integer overflow
Used in MatSeqXAIJSetPreallocation() to determine if the number of nonzeros in the matrix
(and hence the ai[] column indices and lengths of a
Add PetscIntCast() to check for integer overflow
Used in MatSeqXAIJSetPreallocation() to determine if the number of nonzeros in the matrix
(and hence the ai[] column indices and lengths of aj[] columns indices and a[] numerical values) fit
in PetscInt.
Commit-type: error-checking
Reported-by: Karl Lin <karl.linkui@gmail.com>
show more ...
|
| 6b804ed2 | 30-Jul-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'stefano_zampini/GPU-matdensecuda' [PR #1911]
* stefano_zampini/GPU-matdensecuda:
GPU: Initial implementation for SeqDense class on GPUs. |
| 383c9097 | 29-Jul-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'barry/feature-arch-hash' [PR #1796]
* barry/feature-arch-hash:
Option -arch-hash will generate a PETSC_ARCH based on a hash of the config directory.
Based on a suggestion by Satish Bal
Merge branch 'barry/feature-arch-hash' [PR #1796]
* barry/feature-arch-hash:
Option -arch-hash will generate a PETSC_ARCH based on a hash of the config directory.
Based on a suggestion by Satish Balay.
show more ...
|
| f20a8c50 | 26-Jul-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'karlrupp/remove-ancient-cuda-version-checks' [PR #1909]
* karlrupp/remove-ancient-cuda-version-checks:
Remove checks for ancient CUDA versions |
| b139225a | 24-Jul-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
MATSEQAIJ: minor coding style changes |
| 3cb7dd0e | 26-Jul-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
MATMUMPS: Schur complement is updated on the CPU |
| c1fb3f03 | 24-Jul-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
MATSEQAIJCUSPARSE: optimize write access when possible
In MatMult and MatMultAdd_ with compressed rows, we only write access to the GPU array output |
| 1541652f | 24-Jul-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
MATSEQAIJVIENNACL: minor changes |
| e2cf4d64 | 23-Jul-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
Mat: get rid of the valid_GPU_matrix flag from the interface
A specific implementation may provide those operations on GPU
The previous interface would have invalidated the data
MatAssemblyEnd_MPIA
Mat: get rid of the valid_GPU_matrix flag from the interface
A specific implementation may provide those operations on GPU
The previous interface would have invalidated the data
MatAssemblyEnd_MPIAIJ now takes care of propagating the correct
flags to the blocks if we set its flag to OFFLOAD_CPU
This is to fix the case of MatFDColoring, which do not use the MatSetValues interface
show more ...
|
| 489de41d | 22-Jul-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
MatSEQAIJ{CUSPARSE|VIENNACL}: do not copy to the GPU if not at the final stage of assembly |
| b5bb3eec | 22-Jun-2019 |
Mark Adams <ma2325@columbia.edu> |
added missing settypes |