There was an issue where if VTK-m was compiled with CUDA support but then
run on a computer where no CUDA device was available, an inappropriate
exception was thrown (instead of just disabling the device). The
initialization code should now properly check for the existance of a CUDA
device.
The `ExtractGeometry` filter was outputing datasets containing
`CellSetPermutation` as the representation for the cells. Although this is
technically correct and a very fast implementation, it is essentially
useless. The problem is that any downstream processing will have to know
that the data has a `CellSetPermutation`. None do (because the permutation
can be on any other cell set type, which creates an explosion of possible
cell types).
Like was done with `Threshold` a while ago, this problem is fixed by deep
copying the result into a `CellSetExplicit`. This behavior is consistent
with VTK.
The Intel compiler by default turns on an optimization that assumes that
all floating point values are finite. This breaks any ligitimate uses of
non-finite values including checking values with functions like `isnan`
and `isinf`. Turn off this feature for the intel compiler.
There are some overloads for atomic adds of floating point numbers for
older versions of cuda that do not include them directly. These were
misnamed and thus not properly overloading the generic implementation.
This caused compile errors with older versions of cuda.
23fbedf54 Order the VTK scalar types to favor used types
053237e90 Favor base types used by VTK-m when making new basic array
Acked-by: Kitware Robot <kwrobot@kitware.com>
Acked-by: Li-Ta Lo <ollie@lanl.gov>
Merge-request: !3171
Reorder the `VTKScalarTypes` to put the repeated types at the bottom.
This will avoid using types like `long` that are essentially the same as
other types (e.g. `long long`) but will still be considered different.
This causes problems when checking against some VTK types (such as
`vtkIdType`).
When calling `NewInstanceBasic` on an `UnknownArrayHandle`, all C base
types will be tried. In some cases, multiple type will match the same
array. When this happens, favor types used by VTK-m (e.g. `long long`
over `long`).
The flying edges version of the contouring filter converted the isovalues
provided into the same type as the field. This is fine for a floating point
field, but for an integer field the isovalue was truncated to the nearest
integer.
This is problematic because it is common to provide a fractional isovalue
(usually N + 0.5) for integer fields to avoid degenerate cases of the
contour intersecting vertices. It also means the behavior changes between
an integer type that is directly supported (like a `signed char`) or an
integer type that is not directly supported and converted to a floating
point field (like potentially a `char`).
This change updates the worklets to allow the isovalue to have a different
type than the field and to always use a floating point type for the
isovalue.
Threshold was producing wrong results with options `SetAllInRange(false)` and
`SetComponentToTestToAll` because the logic of running
`worklet::Threshold::RunIncremental` on individual components of the input
field and combining the results is incorrect for this case.
With this fix, component modes 'Any' and 'All' are handled by applying
the threshold criteria to each component of each value of the field,
combining the results, and running the threshold worklet on the result
array.
Previously, Clip was updated to remove unused points from the input.
This requires a re-mapping of point ids after compaction. This
re-mapping was missed for the points used for interpolation of centorid
points.
The flying edges algorithm (used when contouring uniform structured cell
sets) was not interpolating cell fields correctly. There was an indexing
issue where a shortcut in the stepping was not incrementing the cell index.
There was a bug in `CleanGrid` when removing degenerate polygons where it
would not detect if the first and last point were the same. This has been
fixed.
There was also an error with function overloading that was causing 0D and
3D cells to enter the wrong computation for degenerate cells. This has also
been fixed.
Fixes#796
* origin/master: (259 commits)
Adding documentation for flow filter restructure
Changes to address feedback from MR
Correct export in class declaration
Support any cell type in MIR filter
Fixing linking issues for flow Analysis class
Removing unnecessary file
Fixing linking issues for flow Analysis class
Fixing linking issues for flow Analysis class
Adding test dependency of filter_flow on tests
Separating Analysis into header and source
Moving base class flow filters into CXX
Fixing streamline output for no step
Removing particle property num punctures
Removing XGC Filter
Fixing WarpXStreamlines compile problems
Fixing issues after rebasing
Removing cout from device code
Changing solver type for WarpX to Euler
Adding specialization for Particle/Field for XGC
ci: disable UnitTestWorkletParticleAdvection in hip build
...
67b7543a3 Adding documentation for flow filter restructure
dbc873efa Changes to address feedback from MR
67716402b Correct export in class declaration
6d1d4f90a Fixing linking issues for flow Analysis class
adcb42455 Removing unnecessary file
78ca3f301 Fixing linking issues for flow Analysis class
0e1ade83a Fixing linking issues for flow Analysis class
12a3bc94e Adding test dependency of filter_flow on tests
...
Acked-by: Kitware Robot <kwrobot@kitware.com>
Merge-request: !3087