| c1838105 | 05-Mar-2020 |
Jed Brown <jed@jedbrown.org> |
VecCUDA: count bytes using size_t
Modify argument of VecCUDA{Set|Get}PinnedMemoryMin to use size_t.
Document how to set this minimum via options database.
Commit-type: portability-fix
Reported-by:
VecCUDA: count bytes using size_t
Modify argument of VecCUDA{Set|Get}PinnedMemoryMin to use size_t.
Document how to set this minimum via options database.
Commit-type: portability-fix
Reported-by: Richard Tran Mills <rmills@rmills.org>
show more ...
|
| e65beb55 | 05-Mar-2020 |
Richard Tran Mills <rmills@rmills.org> |
Ensure that cudaFreeHost() is invoked when needed in VecGetLocalVector_SeqCUDA(). |
| 85558e85 | 30-Jan-2020 |
Richard Tran Mills <rmills@rmills.org> |
bug fix: Don't access veccuda if it is NULL in VecCUDAAllocateCheckHost(). |
| 6c77e611 | 30-Jan-2020 |
Richard Tran Mills <rmills@rmills.org> |
Fixes so MPICUDA Vecs will use cudaFreeHost() when needed. |
| c203764a | 30-Jan-2020 |
Richard Tran Mills <rmills@rmills.org> |
Bug fix: v->spptr must now be freed *after* VecDestroy_SeqCUDA_Private().
VecDestroy_SeqCUDA_Private() needs to access members of the Vec_CUDA data structure. |
| 57f66e68 | 30-Jan-2020 |
Richard Tran Mills <rmills@rmills.org> |
Add ability to set minimum size for pinned host allocations for VECCUDA from command line.
This minimum size now defaults to zero.
Also fixes a bug: for VECSEQCUDA, this minimum size was not being i
Add ability to set minimum size for pinned host allocations for VECCUDA from command line.
This minimum size now defaults to zero.
Also fixes a bug: for VECSEQCUDA, this minimum size was not being initialized to a default.
show more ...
|
| 8a921008 | 26-Jul-2019 |
Hong Zhang <hongzhang@anl.gov> |
Add API to change threshold for using pinned memory
Funded-by:
Project: PETSc for GPU
Time:
Reported-by:
Thanks-to: |
| 31f06eaa | 26-Jul-2019 |
Hong Zhang <hongzhang@anl.gov> |
Add support for pinned memory and use it on VecCUDA
The pinned memory is used when the requested memory size is over 128MB.
Funded-by:
Project: PETSc for GPU
Time:
Reported-by:
Thanks-to: |
| 6f7e44de | 11-Mar-2020 |
Satish Balay <balay@mcs.anl.gov> |
xlf fixes
- add HAVE_F90_ASSUMED_TYPE_NOT_PTR for xlf
- update PetscSFBcastBegin()/PetscSFBcastEnd() fortran interface to use F90_ASSUMED_TYPE_NOT_PTR [ref: src/vec/is/sf/examples/tutorials/ex1f.F90]
xlf fixes
- add HAVE_F90_ASSUMED_TYPE_NOT_PTR for xlf
- update PetscSFBcastBegin()/PetscSFBcastEnd() fortran interface to use F90_ASSUMED_TYPE_NOT_PTR [ref: src/vec/is/sf/examples/tutorials/ex1f.F90]
- fix write format diffs
- fix F2003 requirement
show more ...
|
| 22a8f86c | 05-Mar-2020 |
Lisandro Dalcin <dalcinl@gmail.com> |
PETSCVIEWERBINARY: Various fixes and enhancements
* MPI-IO support for FILE_MODE_APPEND (write at end of file)
* PetscViewerBinaryWriteReadMPIIO() now does IO only in rank zero
* PetscViewerBinaryAd
PETSCVIEWERBINARY: Various fixes and enhancements
* MPI-IO support for FILE_MODE_APPEND (write at end of file)
* PetscViewerBinaryWriteReadMPIIO() now does IO only in rank zero
* PetscViewerBinaryAddMPIIOOffset() is now logically collective
* Fix PetscViewerSetFromOptions() to print current option values
* Add missing error checking in set/get routines
show more ...
|
| 69e10bba | 05-Mar-2020 |
Lisandro Dalcin <dalcinl@gmail.com> |
MPI-IO: Use explicit offsets for {IS|Vec}{View/Load} binary |
| 92f9df4a | 06-Mar-2020 |
Richard Tran Mills <rmills@rmills.org> |
Rename "pintocpu" in VecOps and MatOps to "bindtocpu".
This change should have been made in the commit that changed XXXPinToCPU()
to XXXBindToCPU(), but it was overlooked. |
| b778fa18 | 05-Mar-2020 |
Valeria Barra <valeriabarra21@gmail.com> |
Section field component names |
| 4cb80407 | 23-Feb-2020 |
Satish Balay <balay@mcs.anl.gov> |
Merge remote-tracking branch 'origin/maint' |
| 23cde526 | 22-Feb-2020 |
Satish Balay <balay@mcs.anl.gov> |
cuda: fix cxx build issue with VecGetArrayWrite_SeqCUDA
/people/abhy245/software/petsc/debug-mode-newell/lib/libpetsc.so: undefined reference to `VecGetArrayWrite_SeqCUDA'
Reported-by: "Abhyankar,
cuda: fix cxx build issue with VecGetArrayWrite_SeqCUDA
/people/abhy245/software/petsc/debug-mode-newell/lib/libpetsc.so: undefined reference to `VecGetArrayWrite_SeqCUDA'
Reported-by: "Abhyankar, Shrirang G" <shrirang.abhyankar@pnnl.gov>
show more ...
|
| f0a86a08 | 12-Feb-2020 |
Barry Smith <bsmith@mcs.anl.gov> |
Add missing Fortran interface files
Commit-type: feature, example
Reported-by: "David, Cedric H (US 329F)" <cedric.david@jpl.nasa.gov> |
| 19caf8f3 | 21-Feb-2020 |
Satish Balay <balay@mcs.anl.gov> |
Remove usage of PETSC_HAVE_FORTRAN_STDCALL, PETSC_BLASLAPACK_STDCALL, HAVE_FORTRAN_MIXED_STR_ARG flags - as Compaq f90 compiler is no longer supported |
| b9d19a4e | 21-Feb-2020 |
Satish Balay <balay@mcs.anl.gov> |
Merge remote-tracking branch 'origin/maint' |
| c5062165 | 20-Feb-2020 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'barry/2020-02-15/fix-hdf-string-attribute/maint' into 'maint'
PetscViewerHDF5ReadAttribute() did not handle strings correctly
See merge request petsc/petsc!2532 |
| 245d9833 | 20-Feb-2020 |
prj- <pierre.jolivet@enseeiht.fr> |
Typos in documentation |
| 000b0a7d | 18-Feb-2020 |
Satish Balay <balay@mcs.anl.gov> |
Merge branch 'jed/fix-dmplexclosureindices-local-section' into 'master'
DMPlexGetClosureIndices: fix indexing for local section
See merge request petsc/petsc!2526 |
| 9759eb26 | 17-Feb-2020 |
Jed Brown <jed@jedbrown.org> |
PetscSectionSetConstraintIndices: clarify sorted property of array |
| c7b73be9 | 17-Feb-2020 |
Vaclav Hapla <vaclav.hapla@erdw.ethz.ch> |
vec_vec_tests-ex48: remove - from output |
| bd08f1cb | 17-Feb-2020 |
Vaclav Hapla <vaclav.hapla@erdw.ethz.ch> |
vec_vec_tests-ex48 can now test string attributes |
| 708d4cb3 | 16-Feb-2020 |
Barry Smith <bsmith@mcs.anl.gov> |
PetscViewerHDF5ReadAttribute() did not handle strings correctly
Fixes provided by Carl Hall <carl.a.hall@vanderbilt.edu>
Commit-type: bug-fix
Reported-by: Carl Hall <carl.a.hall@vanderbilt.edu> |