Mainly issue dealing with dimensionality of cell sets and what that represents.
Have added in code to allow user to specify a custom dimensionality so that
tests continue to work properly.

If you have only a single cell shape, you can now efficiently create
a vtkm::cont::DataSet whose CellSet is vtkm::cont::CellSetSingleType.

Robert Maynard authored Jan 15, 2016 and Kitware Robot committed Jan 15, 2016

956cedfd Turn off the benchmarking ExternalsFaces.
18b866d6 Threshold worklet is not templated on device adapter.
dbee9275

 ExternalFaces worklet is not templated on device adapter.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !315

Synchronize the CUDA timer on both the start and end events

Previously, the timer for CUDA devices only called cudaEventSynchronize
at the end event when asking for the elapsed time. This, however, could
allow time to pass from when the timer was reset to when the start event
happened that was not recorded in the timer. This added synchronization
should make sure that all time spent in CUDA is recorded.

See merge request !291

DataSetBuilder fixes

There were some compiler warnings and testing issues with the
recently merged DataSetBuilder branch. This fixes some of
those problems.

See merge request !316

Reduce the number of conditions UnitTestArrayHandleCartesianProduct
tries so that it is faster. All the conditions are pretty similar, so it
should be OK to reduce some.

Previously UnitTestDataSetBuilderRectilinear and
UnitTestDataSetBuilderRegular did an exhaustive test of all possible
grid sizes within a certain amount of dimensions and with some number of
ways to build point arrays. This created hundreds of thousands of cases
that were checked, and it was running a long time. At best it wasted
time and at worst ctest reported the test as timed out.

It is not really necessary to perform an exhaustive test. The tests are
changed to do 10 trials using random values for dimensions and fill
methods.

Robert Maynard authored Jan 14, 2016 and Kitware Robot committed Jan 14, 2016

12ddcfdd

 VertexClustering worklet is not templated on device adapter.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !314

This should help reduce the amount of code generation, when building the
Threshold worklet for all device adapters.

This should help reduce the amount of code generation, when building the
ExternalFaces worklet for all device adapters.

This should help reduce the amount of code generation, when building the
VertexClustering worklet for all device adapters.

Dave Pugmire authored Jan 14, 2016 and Kitware Robot committed Jan 14, 2016

f86382f0 Fix support for CoordinateSystems using ArrayHandleCartesianProduct.
d6a2a142 Add toleranced compare for values. Add tests for vtkm::Float32,Float64,Id typed arrays.
5d438353 Add toleranced comparisions for bounds validation. Also, add vtkm::Float32 and vtkm::Float64 to the testing for rectilinear and regular datasets.
b225ae97 Rectilinear coordinates (created with DataSetBuilderRectilinear) are now converted to vtkm::FloatDefault. This reduces the number of types to consider when casting inside CoordinateSystem, and was felt by all to be a reasonable restriction.
d755e43d Use ArrayHandleCompositeVector to represent separated point arrays for DataSetBuilderExplicit.h.
c7b0ffb8 Add tests for DataSetBuilderExplicit. Added cont/testing/ExplicitTestData.h which includes several explicit datasets.  These datasets come from VTK data generated in VisIt.  The new unit tests build datasets in several different ways and do some basic validation.
b4d04fff Add specialization of printSummary_ArrayHandle for UInt8. It prints them as characters, which are a little hard to understand to this computer scientist.
bd929c20

 Fix compiler warnings.
...
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Kenneth Moreland <kmorel@sandia.gov>
Merge-request: !262

Robert Maynard authored Jan 14, 2016 and Kitware Robot committed Jan 14, 2016

4e94617f

 Make DynamicCellSet use VTKM_DEFAULT_CELL_SET_LIST_TAG.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !309

Robert Maynard authored Jan 13, 2016 and Kitware Robot committed Jan 13, 2016

15e1f80d

 Generalize MarchingCubes input with additional template parameters.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Robert Maynard <robert.maynard@kitware.com>
Merge-request: !304

Robert Maynard authored Jan 13, 2016 and Kitware Robot committed Jan 13, 2016

f5f9939f Update all of vtkm to understand it can only identify as one compiler.
c706c826

 Configure.h can only state a machine is a one compiler.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !307

Robert Maynard authored Jan 13, 2016 and Kitware Robot committed Jan 13, 2016

b8e5923a

 FunctorsTBB now uses VTKM_VECTORIZATION_X macros everywhere.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Chuck Atkins <chuck.atkins@kitware.com>
Merge-request: !306

It was previously using vtkm::cont::CellSetListTagCommon which would
break the ability for people to specify a custom value for
VTKM_DEFAULT_CELL_SET_LIST_TAG.

Add lots of checks to CUDA calls in the timer to try to identify any
problems that might be showing up on the dashboard.

Also adding some print statements around the sleep function in the
device adapter testing. For some reason the problem just went away with
them.

Previously, the timer for CUDA devices only called cudaEventSynchronize
at the end event when asking for the elapsed time. This, however, could
allow time to pass from when the timer was reset to when the start event
happened that was not recorded in the timer. This added synchronization
should make sure that all time spent in CUDA is recorded.

Chuck Atkins authored Jan 12, 2016 and Kitware Robot committed Jan 12, 2016

ccdb7bb6

 Fix incorrect vectorization pragma for GCC
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !308

It used to be possible for vtk-m to say it was multiple compilers, for example
it could be both GCC and PGI, Clang and MSVC ( yes possible ), or Intel and
Clang.

This logical restructure now makes that impossible, and instead prefers a system
where we choose the most specialized version of the compiler over the most
general, where general is GCC / Clang.

Robert Maynard authored Jan 11, 2016 and Kitware Robot committed Jan 11, 2016

9e9e3caf

 Fix a misplaced '\' for GCC pragmas in Configure.h.in
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Robert Maynard <robert.maynard@kitware.com>
Merge-request: !305

Robert Maynard authored Jan 08, 2016 and Kitware Robot committed Jan 08, 2016

b70a000a

 Allow resetting the Type and Storage of a DynamicArrayHandle in a single call.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Kenneth Moreland <kmorel@sandia.gov>
Merge-request: !302

Robert Maynard authored Jan 08, 2016 and Kitware Robot committed Jan 08, 2016

82a573f7 MarchingCubes is now able to not generate normals.
502e7c28 Merge branch 'cleanup_scatter_counting_uses' into marching_cubes_normal_generation_option
8079dc28 MarchingCubes generate step now requires a ScatterCounting object.
4cd2f582

 Add a default constructor for ScatterCounting.
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Kenneth Moreland <kmorel@sandia.gov>
Merge-request: !300

vtkm::filter has the use case where we need to reset both the type and
storage of an array handle, by doing both at the same time we can reduce
the number of temporary objects, and invalid conversions of arrays.

Add PointCount to WorkletMapPointToCell.

WorkletMapPointToCell is a convenience subclass of WorkletMapTopology.
As such, it renames all the From/To signature tags to say Point/Cell to
be easier to read. However, the alias for FromCount was missing. Add the
alias PointCount.

See merge request !303

WorkletMapPointToCell is a convenience subclass of WorkletMapTopology.
As such, it renames all the From/To signature tags to say Point/Cell to
be easier to read. However, the alias for FromCount was missing. Add the
alias PointCount.

Add tests for vtkm::Float32,Float64,Id typed arrays.

Also, add vtkm::Float32 and vtkm::Float64 to the testing for rectilinear and regular datasets.

Rectilinear coordinates (created with DataSetBuilderRectilinear) are now converted to vtkm::FloatDefault.
This reduces the number of types to consider when casting inside CoordinateSystem, and was felt by all to be a reasonable restriction.