Robert Maynard
0748cdb320
Merge branch 'implicit_shortening_on_mult'
2015-06-03 08:46:30 -04:00
Robert Maynard
d15380b4ca
Merge branch 'suppress_warnings_in_tbb'
2015-06-03 08:46:22 -04:00
Robert Maynard
e0e6ffb705
Suppress warnings generated by TBB code, which we can change.
...
These warnings are caused by tbb code, which we have zero control over.
2015-06-02 09:42:40 -04:00
Robert Maynard
f427d6d8d8
Cast back to int8/16 after compiler does integer promotion.
...
When multiplying integer values of a width less than an integer c++ actually
converts up to an integer implicitly. So we silence the warnings that the
result of the multiply could be larger than the original type. If people want
to do multiplies on int8/16 they better know what they are doing.
2015-06-02 09:10:46 -04:00
Robert Maynard
9332becc47
Merge branch 'correct_vec_constructor'
2015-06-01 17:20:34 -04:00
Robert Maynard
f88992e2f7
cuda can't handle invoking worklets from anonymous namespaces.
2015-06-01 17:20:02 -04:00
Robert Maynard
3a532160c9
cuda device adapter now support different value types for keys and values.
2015-06-01 17:19:33 -04:00
Robert Maynard
d2b8e922a2
Merge branch 'correct_vec_constructor'
2015-06-01 14:51:22 -04:00
Robert Maynard
98bd557a4f
TestingDeviceAdapter now doesn't use initializer lists for vtkm::Vec
...
Since initializer lists are a c++11 feature we shouldn't use them.
2015-06-01 14:50:35 -04:00
Robert Maynard
ca3a91ef6e
Merge branch 'add-tbb-backend'
2015-06-01 14:31:01 -04:00
Sujin Philip
08f88b1cb9
Add TBB backend.
2015-06-01 13:57:37 -04:00
Robert Maynard
c2ff4d5870
Merge branch 'correct_reduce_by_key_serial'
2015-06-01 12:33:31 -04:00
Chun-Ming Chen
7ea9563d19
Fixed bug in ReduceByKey
2015-06-01 12:33:08 -04:00
Robert Maynard
b695460f39
Merge branch 'faster_extent_methods'
2015-05-28 09:12:04 -04:00
Robert Maynard
6910e70bca
Merge branch 'signed_conversion_warnings'
2015-05-28 09:10:23 -04:00
Robert Maynard
07970cf476
Correct all signed / unsigned and narrowing warnings ( 64bit to 32bit ).
2015-05-28 09:05:17 -04:00
Robert Maynard
100b894c13
Enable signed / unsigned and 64bit/32bit conversion warnings.
2015-05-28 09:05:16 -04:00
Robert Maynard
e17b072ac6
Add optimized versions of ExtentIndexToFlatIndex for dim=1,2,3.
2015-05-28 08:28:13 -04:00
Robert Maynard
e09b936b23
Merge branch 'blessley/vtk-m-impl_conv_changes'
2015-05-26 09:35:03 -04:00
Brent Lessley
0a72789304
Resolved all implicit conversions between unsigned int and vtkm::Id.
2015-05-26 09:34:43 -04:00
Robert Maynard
c16d6d93e8
Merge branch 'blessley/vtk-m-fp_comp_warning'
2015-05-26 09:19:38 -04:00
Brent Lessley
2a89f66144
Resolved all compiler warnings for floating-point comparisons in TestingDeviceAdapter.h
2015-05-26 09:18:19 -04:00
Robert Maynard
d54aee7eb5
Merge branch 'fix_typo_in_copyright'
2015-05-21 10:32:08 -04:00
Robert Maynard
6b8e7822be
The Copyright statement now has all the periods in the correct location.
2015-05-21 10:30:11 -04:00
Robert Maynard
38258e0f71
Merge branch 'reduce_by_key'
2015-05-21 08:50:39 -04:00
Robert Maynard
52b87b1945
Merge branch 'remove_duplicate_testing_code'
2015-05-21 08:50:29 -04:00
Robert Maynard
0dde5fec68
Remove duplicate testing code from UnitTestFunctionInterface.
2015-05-21 08:50:08 -04:00
Robert Maynard
f2b47ffd4a
Add ReduceByKey to the DeviceAdapter.
2015-05-21 08:23:28 -04:00
Robert Maynard
be193542ac
Introduce StreamCompact with a custom stencil unary predicate.
2015-05-21 08:23:28 -04:00
Robert Maynard
132fa7e1e5
make vtkm::not_default_constructor operator const.
2015-05-20 11:16:49 -04:00
Robert Maynard
959d89f77d
Correct DeviceAdapterAlgorithmGeneral ScanInclusive with length 1.
...
Previously we returned a sum of 0 when the length is 1, when really
we need to return the value that is in side the single location of the
array.
2015-05-20 09:30:36 -04:00
Robert Maynard
d4adcb464a
Merge branch 'reduce'
2015-05-19 16:53:59 -04:00
Kenneth Moreland
f514a9c72a
Merge branch 'msvc-problems'
2015-05-18 07:39:17 -06:00
Kenneth Moreland
0011e9d809
Check that appropriate exception thrown in UnitTestDispatcherBase.
2015-05-18 07:38:52 -06:00
Kenneth Moreland
4276b0cdcb
Fix warning about uninitialized object
...
It is not important for this test, but we might as well make the
compiler happy.
2015-05-17 18:32:07 -06:00
Kenneth Moreland
97847fb9f4
Fix warning about converting between an int and a float.
2015-05-17 18:29:35 -06:00
Robert Maynard
e38caafe37
Adding Reduce with custom operator to the DeviceAdapterAlgorithm.
2015-05-14 15:16:49 -04:00
Robert Maynard
5d9f369d0c
Adding ScanInclusive with custom binary operator to DeviceAdapterAlgorithm.
2015-05-14 15:16:49 -04:00
Robert Maynard
9519737b3a
Adding Reduce to the DeviceAdapterAlgorithm.
2015-05-14 15:16:49 -04:00
Robert Maynard
fbee20fc03
Mark the local value for the result of GetNumberOfValues as const.
...
We aren't modifying the value, so no reason to not have it as const.
2015-05-14 15:16:48 -04:00
Kenneth Moreland
84ea9b8490
Merge branch 'exec-object-fix'
2015-05-14 10:20:20 -06:00
Kenneth Moreland
c83b585b14
Fix ExecObject tag in WorkletBase
...
The ExecObject tag for the ControlSignature was not declared right so
would cause a compile error if it was ever used. Clearly this was not
being tested properly, so the dispatcher base unit test now passes an
ExecObject parameter.
2015-05-13 18:44:44 -06:00
Kenneth Moreland
4afb6fe5ff
Merge branch 'constant-array-handle'
2015-05-13 14:25:23 -06:00
Kenneth Moreland
d77342a7c6
Add ArrayHandleConstant
2015-05-13 14:24:51 -06:00
Robert Maynard
59618b3a50
Merge branch 'sort_by_key'
2015-05-11 08:04:09 -04:00
Robert Maynard
ca0d1ade55
Merge branch 'test_permutation_as_output'
2015-05-08 15:32:37 -04:00
Robert Maynard
1bafa94194
Verify the permutation handle can be used as output on all devices
2015-05-08 15:32:06 -04:00
Kenneth Moreland
9db8483b5a
Merge branch 'msvc-warnings'
2015-05-08 10:55:24 -06:00
Kenneth Moreland
dd897c0565
Remove template parameter for IncrementBy2
...
THe IncrementBy2 test type previously allowed any subtype including
floating point numbers. The meaning of this is actually a little unclear
and the feature was causing implicit type conversion warnings that were
hard to template out. The utility of of templating this class is dubious
in the first place, so class is now a fixed type.
I'm a little unsure whether we should keep this test class at all. It's
math operations are ad hoc and it could be difficult to determine if a
problem is caused by an actual problem or just bad math operators.
2015-05-08 10:53:28 -06:00
Kenneth Moreland
209053b79e
Fix issues with implicitly converting ints to floats
...
MSVC complains that converting a 32-bit int to a 32-bit float could
cause loss of precision.
2015-05-08 10:33:59 -06:00