Merge branch 'madams/gamg-destroy' into prbrune/pcgamg-classicalinterpolationstrategies

Merge branch 'master' into madams/gamg-destroy

Conflicts:
src/ksp/pc/impls/gamg/gamg.c

Merge branch 'master' into knepley/feature-plex-refine-hex

* master: (68 commits)
DMPlex: More fixup from bad rebase, moved to DMGet/SetCoordinateSection()
DM: Added DMGet/SetCoordinateSection()

Merge branch 'master' into knepley/feature-plex-refine-hex

* master: (68 commits)
DMPlex: More fixup from bad rebase, moved to DMGet/SetCoordinateSection()
DM: Added DMGet/SetCoordinateSection()
Fuckup: Fix for merge that is still untangling bad rebase from Oct. 17th 2013 - Fixed const in declarations - Fixed merge from PetscFE - Fixed 2 mallocs
DMDA: Missing header - Damn rebase
DMPlex ex3: Added 2D Q_1 and 3D P_2 tests
DMDA: Added DMDAProjectFunction() and DMDAComputeL2Diff()
DMDA: Fixed 2D geometry - Should reuse DMPlex routines
DMDA: Fix bug in 2D closure operation for cells
DMDA: Make closure operations also return the size
DMDA: Changed DMDACreateSection() to emulate DMPlexCreateSection()
DMDA: Added functions which emulate DMPlex functionality
DMDA: Now DMDAGetNumCells() returns the cells in each direction as well
PetscFE: Added support for tensor product cells in PetscDualSpace_Lagrange
PetscFE: Added tensor product polynomial spaces
example fix /sandbox/petsc/petsc.clone-3/include/petscmath.h(260): error: identifier "PETSC_CXX_STATIC_INLINE" is undefined PETSC_STATIC_INLINE PetscReal PetscAbsScalar(PetscScalar a) {return a < 0.0 ? -a : a;} ^
remove warning in fun3d example "user.F", line 1256: warning: In-place macro substitution leaves line truncated "user.F", line 1259: warning: In-place macro substitution leaves line truncated "user.F", line 1262: warning: In-place macro substitution leaves line truncated
/Users/petsc/petsc.clone-2/src/ksp/ksp/examples/tutorials/ex56.c:59: warning: comparison between signed and unsigned integer expressions
fix examples for portability
fixed to example outputs
fix output for a few examples changed due to changes to -mat_view and not using -snes_monitor_short
...

Conflicts:
src/dm/impls/plex/plexrefine.c

show more ...

Merge branch 'barry/update-xxxviewfromoptions'

Merge branch 'master' into barry/xcode

VecAssembly BTS: VEC_SUBSET_OFF_PROC_ENTRIES elides global synchronization

SUBSET is currently a misnomer because the entries must currently match
exactly. That will be fixed by checking an MPI_Sta

VecAssembly BTS: VEC_SUBSET_OFF_PROC_ENTRIES elides global synchronization

SUBSET is currently a misnomer because the entries must currently match
exactly. That will be fixed by checking an MPI_Status.

show more ...

introduce XXXViewFromOptions() and use consistently when possible

Merge branch 'master' into knepley/feature-fem-dgspace

* master: (593 commits)
Bib: Added Top500 and fixed entry
fun3d: update PetscMallocValidate() usage
bib: rename ref with duplicate key

Merge branch 'master' into knepley/feature-fem-dgspace

* master: (593 commits)
Bib: Added Top500 and fixed entry
fun3d: update PetscMallocValidate() usage
bib: rename ref with duplicate key
Webpage: Corrected '-dm_mat_type cusp' to 'dm_mat_type aijcusp'.
Allow calling MatGetBlockSize[s]() before matrix preallocation
SNESLINESEARCHBT: Set the norms when exiting early due to negligible step.
SNESQN: only monitor real part of dot product (fails with C++ complex)
Corrected the #include statement in the man pages for the PetscObjectComposedDataSet/Getwhatever routines.
DMPlex: Added doc for DMPlexGetHybridBounds()
winzip: detect if winzip is used to extract petsc.tar.gz and error out.
configure: remove self.archIndependent as its not being used. [it is used to create externalpackages/package/arch dir which is unused]
configure: fix packages using package-dir/PETSC_ARCH as out-of-source-build location Perhaps this should be changed to package-dir/build?
configure: remove dead code previously used to download BuildSystem
configure: check if compilerDefines [and compilerFixes] exist in framework before using them
configure: save/restore reconfigure.py when --with-clean is used
configure: add --with-clean option to delete buildfiles/externalpackages in PETSC_DIR/PETSC_ARCH and potential externalpackages in --with-external-package-dir/PETSC_ARCH
configure: With --with-externalpackages-dir=dir store/build packages in dir/PETSC_ARCH
Add MatSeqSBAIJSetPreallocationCSR()
configure: switch to using PETSC_DIR/PETSC_ARCH/externalpackages by updating package.py to use externalpackages.py as the externalPackagesDirProvider
configure: move dead code configureExternalPackagesDir() from petscdir.py to externalpackagesdir.py Also set PETSC_DIR/PETSC_ARCH/externalpackages as the externalpackagesdir
...

show more ...

Merge branch 'prbrune/sf-sfbasicops' into prbrune/mat-matcolor

Merge branch 'master' into barry/saws

Conflicts:
src/ksp/pc/impls/gamg/gamg.c
src/sys/classes/viewer/impls/ams/ams.c
src/sys/objects/pinit.c

Merge branch 'master' into knepley/feature-dmda-section

* master: (397 commits)
PetscSynchronizedFGets: fix deadlock at EOF
Compiler: Fix warnings from MPI impls which do not initialize outputs

Merge branch 'master' into knepley/feature-dmda-section

* master: (397 commits)
PetscSynchronizedFGets: fix deadlock at EOF
Compiler: Fix warnings from MPI impls which do not initialize outputs
Increase patchlevel to 3.4.3
SNES: Now work vectors come from the DM
SNES ex62: Remove code generation
SNES: Move setup involving snes->vec_sol from SNESSetUp() to SNESolve()
Compiler: Fix warnings from MPI impls which do not initialize outputs
DMPlex ex7: Added missing test output
SNES ex52: Removed old Jacobian stuff, and fixed call to DMPlexProjectFunction() - Fixed calls for new element handling
SNES ex12: Added a performance profiling mode
DMPlex ex8: Fixed leak
PetscSection: Must reset the section when changing the number of fields
DMPlex ex7: Fixed test output - Corrected orientations in interpolation
DMPlex ex1: Fixed test output - Uniform refinement for quads changed
DMPlex: Fix overagressive checks
PC: Removed support graph PC
configure: both downloadonWindows and worksonWindows refer to MS compilers. Also fix isWindows() -> isWindows(CC).
MatXAIJSetPreallocation: use array[] notation to help out Fortran
__float128: when blas/lapack is not found instruct using f2cblaslapack
Configure: Package.downloadonWindows is supposed to mean that Windows compilers work. not Cygwin - Also changed doc for Package.worksonWindows, which does refer to Cygwin
...

Conflicts:
src/dm/impls/da/dalocal.c

show more ...

Merge branch 'knepley/fix-plex-examples' into knepley/feature-plex-refine-3d

* knepley/fix-plex-examples: (36 commits)
DMPlex ex7: Added missing test output
SNES ex52: Removed old Jacobian stuff

Merge branch 'knepley/fix-plex-examples' into knepley/feature-plex-refine-3d

* knepley/fix-plex-examples: (36 commits)
DMPlex ex7: Added missing test output
SNES ex52: Removed old Jacobian stuff, and fixed call to DMPlexProjectFunction() - Fixed calls for new element handling
SNES ex12: Added a performance profiling mode
DMPlex ex8: Fixed leak
PetscSection: Must reset the section when changing the number of fields
DMPlex ex7: Fixed test output - Corrected orientations in interpolation
DMPlex ex1: Fixed test output - Uniform refinement for quads changed
DMPlex: Fix overagressive checks
configure: both downloadonWindows and worksonWindows refer to MS compilers. Also fix isWindows() -> isWindows(CC).
MatXAIJSetPreallocation: use array[] notation to help out Fortran
__float128: when blas/lapack is not found instruct using f2cblaslapack
Configure: Package.downloadonWindows is supposed to mean that Windows compilers work. not Cygwin - Also changed doc for Package.worksonWindows, which does refer to Cygwin
SNES example: fix bad merge
config: only define DATAFILESPATH for non-null value
config: define lang-specific macros in petscconf.h, choose in petscsys.h
Mat: add MATOP_ values for missing functions
DMPlex: use VECSTANDARD for coordinates
PetscSF: Fixed PetscSFCreateEmbeddedSF() - We were using sf->nleaves for the leaf buffer, which is completely wrong. We have to use the largest thing in sf->mine[]
replaced ISCreateGeneral in CompositeDM with ISCreateStride. MatNest calls ISStrideGetInfo on these vectors, which was causing errors
DMPlex: Fix completely broken code in PetscSFCreateRemoteOffsets() - Non-broken example was in PetscSFDistributeSection()
...

Conflicts:
src/dm/impls/plex/plex.c

show more ...

Merge branch 'paulmullowney/cusp-vector-scatter-with-fix'

* paulmullowney/cusp-vector-scatter-with-fix:
CUDA/CUSP: Implementation of Sequential to Sequential Vector Scatters

Original commit by Paul Mullowney (augmented by PETSC_HAVE_CUSP include
guards by Karl Rupp):
Implementation of Sequential to Sequential VecScatters on the GPU.

In this commit, I've built a working

Original commit by Paul Mullowney (augmented by PETSC_HAVE_CUSP include
guards by Karl Rupp):
Implementation of Sequential to Sequential VecScatters on the GPU.

In this commit, I've built a working prototype for sequential
to sequential vector scatters for CUSP vectors. I've also reorganized
the parallel to parallel vector scatters in a new infrastructure.

The design of the code is as follows. Currently, I distinguish
between PtoP (parallel to parallel) VecScatters and StoS (sequential
to sequential) VecScatters. In cuspvecimpl.h, a high level struct
called PetscCUSPIndices stores a void * pointer and enumerated type
describing whether the scatter is PtoP or StoS (and later PtoS and
perhaps StoP). The actual type of scatter information built
and stored in the void * pointer depends on the calling code.

For instance, in vscat.c, there are the methods VecScatterBegin_SGToSG,
VecScatterBegin_SGToSS, ... These methods build the indices for
the sequential scatter on demand via the function
VecScatterCUSPIndicesCreate_StoS. Those indices are stored in
a struct of type _p_VecScatterCUSPIndices_StoS. the appropriate
data in that struct is filled depending on whether the input
and output are general or strided. Some additional meta data is
also stored.

A similar routine is called from vpscat.c for building the
PtoP vector scatter. The structure of this hasn't changed although
the constructing API method is now VecScatterCUSPIndicesCreate_PtoP
(it used to be PetscCUSPIndicesCreate).

A single destructor function is given for the PtoP and StoS scatters:
VecScatterCUSPIndicesDestroy. It used to be PetscCUSPIndicesDestroy.

I've made all of these methods PETSC_INTERN. I can't see any
reason to expose these methods to the user.

All of the source code for doing the sequential scatter computation
on the GPU is moved into a file vecscattercusp.cu (src/vec/vec/impls/seq/seqcusp).
I haven't moved the VecCUSPCopyToGPUSome and similar methods into
this file, although perhaps they belong there.

All scatter types implemented including FORWARD and REVERSE as well
as INSERT, ADD, MAX.

show more ...

Merge branch 'master' into barry/wirth-fusion-materials

Conflicts:
src/ts/examples/tutorials/advection-diffusion-reaction/ex10.c

Merge branch 'barry/dmvecmattypes' into barry/saws

Needed to work with version of PETSc that did not have constant calls to VecSetFromOptions() etc

Conflicts:
src/ksp/ksp/interface/ams/kspams.c
s

Merge branch 'barry/dmvecmattypes' into barry/saws

Needed to work with version of PETSc that did not have constant calls to VecSetFromOptions() etc

Conflicts:
src/ksp/ksp/interface/ams/kspams.c
src/snes/impls/composite/snescomposite.c
src/snes/impls/gs/snesgs.c
src/snes/impls/nasm/nasm.c
src/snes/impls/ngmres/snesngmres.c

show more ...

Merge branch 'master' into knepley/feature-dt-fem

* master: (211 commits)
Mat ex170: Comments
VTK: Small fix to error message (.vts to .vtu)
VTK: Small fix to error message
Fixed bib entries

Merge branch 'master' into knepley/feature-dt-fem

* master: (211 commits)
Mat ex170: Comments
VTK: Small fix to error message (.vts to .vtu)
VTK: Small fix to error message
Fixed bib entries
Bib: Updates
AO: fix erroneous processing of -ao_view and factor into AOViewFromOptions
doc: fix named argument in {Vec,Mat,DM}ViewFromOptions
Sys: add PetscDataTypeFromString() and test code
Mat: Should say that it has a nullspace in MatView()
parms: update tarball with fix for namespace conflict with metis
fix citation 'Golub_Varga_1961'
parmetis: update tarball to parmetis-4.0.2-p5 which fixes an install issue with cygwin
Sys Logging: revert parent traversal
fixed hdf5.py so that if self.libraries.compression is None the code still runs correctly
DMDA: fix bad cast of DM_DA to PetscObject
MatClique: follow DistMultiVec API changes
MatClique: remove unused variables
config cmakeboot: add C++ flags any time compiler is available
config OpenMP: check for C++ flag any time the compiler is available
replaced all left-over uses of a single PetscMalloc() to allocated multiple arrays: replaced with PetscMallocN() The only ones left are when the second array is set into the first array and one ugly usage in the MUMPS interface that cannot be easily fixed
...

Conflicts:
config/builder.py
src/dm/impls/plex/plexgeometry.c

show more ...

Merge branch 'master' into knepley/feature-plex-refine-3d

* master: (273 commits)
Mat ex170: Comments
VTK: Small fix to error message (.vts to .vtu)
VTK: Small fix to error message
Fixed bib

Merge branch 'master' into knepley/feature-plex-refine-3d

* master: (273 commits)
Mat ex170: Comments
VTK: Small fix to error message (.vts to .vtu)
VTK: Small fix to error message
Fixed bib entries
Bib: Updates
AO: fix erroneous processing of -ao_view and factor into AOViewFromOptions
doc: fix named argument in {Vec,Mat,DM}ViewFromOptions
Sys: add PetscDataTypeFromString() and test code
Mat: Should say that it has a nullspace in MatView()
parms: update tarball with fix for namespace conflict with metis
fix citation 'Golub_Varga_1961'
parmetis: update tarball to parmetis-4.0.2-p5 which fixes an install issue with cygwin
Sys Logging: revert parent traversal
fixed hdf5.py so that if self.libraries.compression is None the code still runs correctly
DMDA: fix bad cast of DM_DA to PetscObject
MatClique: follow DistMultiVec API changes
MatClique: remove unused variables
config cmakeboot: add C++ flags any time compiler is available
config OpenMP: check for C++ flag any time the compiler is available
replaced all left-over uses of a single PetscMalloc() to allocated multiple arrays: replaced with PetscMallocN() The only ones left are when the second array is set into the first array and one ugly usage in the MUMPS interface that cannot be easily fixed
...

Conflicts:
include/petscdmplex.h

show more ...

Merge branch 'master' into knepley/feature-dmda-section

* master: (287 commits)
Mat ex170: Comments
VTK: Small fix to error message (.vts to .vtu)
VTK: Small fix to error message
Fixed bib e

Merge branch 'master' into knepley/feature-dmda-section

* master: (287 commits)
Mat ex170: Comments
VTK: Small fix to error message (.vts to .vtu)
VTK: Small fix to error message
Fixed bib entries
Bib: Updates
AO: fix erroneous processing of -ao_view and factor into AOViewFromOptions
doc: fix named argument in {Vec,Mat,DM}ViewFromOptions
Sys: add PetscDataTypeFromString() and test code
Mat: Should say that it has a nullspace in MatView()
parms: update tarball with fix for namespace conflict with metis
fix citation 'Golub_Varga_1961'
parmetis: update tarball to parmetis-4.0.2-p5 which fixes an install issue with cygwin
Sys Logging: revert parent traversal
fixed hdf5.py so that if self.libraries.compression is None the code still runs correctly
DMDA: fix bad cast of DM_DA to PetscObject
MatClique: follow DistMultiVec API changes
MatClique: remove unused variables
config cmakeboot: add C++ flags any time compiler is available
config OpenMP: check for C++ flag any time the compiler is available
replaced all left-over uses of a single PetscMalloc() to allocated multiple arrays: replaced with PetscMallocN() The only ones left are when the second array is set into the first array and one ugly usage in the MUMPS interface that cannot be easily fixed
...

show more ...

Merge branch 'knepley/reordering'

* knepley/reordering:
Mat ex170: Comments
Mat ex170: Test for MatMult() using max instead of plus - Finds the number of connected components in parallel - Can s

Merge branch 'knepley/reordering'

* knepley/reordering:
Mat ex170: Comments
Mat ex170: Test for MatMult() using max instead of plus - Finds the number of connected components in parallel - Can still optimize better in parallel
Mat: Added stuff to let me do (max, mult) algebra things for reordering - Added MatMultMax_SeqAIJ() and MatMultAddMax_SeqAIJ() - Added PetscSparseDenseMaxDot()
Vec: Added VecUniqueEntries() - This is purely diagnostic, but I think its useful for tests
Sys: Added PetscSortRemoveDupsReal()

Conflicts:
config/builder.py

show more ...

Merge remote-tracking branch 'origin/master' into stefano_zampini/pcbddc-improvelocalsolvers

Merge remote-tracking branch 'origin/master' into stefano_zampini/pcbddc-mirrorsfix

Merge remote-tracking branch 'origin/master' into stefano_zampini/pcbddc-constraintssetupimproved

Merge branch 'paulmullowney/txpetscgpu-package-removal'

Removing TXPETSCGPU from veccusp and mpiaijcusparse

In this next step of removing TXPETSCGPU, the host-device and
device-host messaging code has been significantly simplified. In
particular, all met

Removing TXPETSCGPU from veccusp and mpiaijcusparse

In this next step of removing TXPETSCGPU, the host-device and
device-host messaging code has been significantly simplified. In
particular, all methods VecCUSPCopyToGPU/FromGPU now use
a cudaMemcpyAsync with a stream (and a stream synchronize()).
This never hurts you. Moreover, it can help you in the case
of the multi-GPU SpMV as this data transfer will overlap
with the MatMult kernel. The more signficant change comes in
VecCUSPCopyToGPUSome and VecCUSPCopyFromGPUSome. In this code,
the data transfer now moves the smallest contiguous set of
vector data containing ALL the indices in a single asynchronous data
transfer. Then, the stream containing the data transfer is
synchronized (not the entire device). While this can be wasteful
in terms of messaging too much data, it has shown the best
scalability performance across a wide range of matrices. Lastly
the simplicity of the code is a significant advantage over
the old way of doing the data transfer. Some old cold
in these methods is "if 0"-ed out for reference and will be
cleaned up later. One final optimization in the vector code
involves registering the host buffer as page locked--which
is done in VecCUSPAllocateCheck. Then, the buffer must be
unregistered at VecDestroy_SeqCUSP. This shows a nice
speedup in the data transfer for a parallel MatMult.

Also in this commit, I am removing the TXPETSCGPU dependence from
the mpiaijcusparse class--it now depends only on CUDA. In order
for the same stream to be used in the MatMult and MatMultAdd
(necessary for an optimal Multi-GPU SpMV), the stream is built
in the mpiaijcusparse and then passed in the seqaijcusparse data
structure via a new method (MatCUSPARSESetStream). A similar method
is added for the CUSPARSE library handle (context) as I think the
stream needs to be attached to a particular context to work properly.
When running in parallel, multiple GPUs, the references to the handle
in the seqaijcusparse are cleared from the mpiaijcusparse classes with
the method MatCUSPARSEClearHandle. Then, the mpiaijcusparse class
deletes the handle.

One other non-trivial change was made to the seqaijcusparse. The alpha
and beta parameters to the SpMV are now device data which is owned by
the Mat_SEQAIJCUSPARSEMultStruct structure. This enables slightly better
multi-GPU performance as this data does not need to be copied to the
GPU at each kernel launch.

Multi-GPU SpMV now works without TXPETSCGPU and the performance is recovered
as tested on up to 4 GPUs. Code is valgrind clean and cuda-memcheck clean.

Results of tests have been modified to have 1 less digit of precision. This
yields consistent results across different GPUs. Lastly, the parallel test
is set to run on a different matrix (shallow_water1) so that the iteration
actually converges.

show more ...