Commit Graph

7536 Commits

Author SHA1 Message Date
Robert Maynard
fdac208acb use cuda scheduling versus thrust scheduling. 2015-04-03 10:18:05 -04:00
Robert Maynard
b7841c1251 Merge branch 'cuda_dont_pre_clear_vectors' 2015-04-03 10:14:09 -04:00
Dave Pugmire
8d99a7f0d1 Merge branch 'master' into datamodel-design 2015-03-25 09:59:55 -04:00
Dave Pugmire
a4f740b933 comments 2015-03-25 09:57:58 -04:00
Dave Pugmire
8f3ed89e92 comments 2015-03-25 09:57:42 -04:00
Dave Pugmire
f45b3363e1 comments 2015-03-25 09:56:48 -04:00
Robert Maynard
d340a573d9 don't clear ::thrust::vectors as that is not our intended behavior. 2015-03-23 09:01:42 -04:00
Robert Maynard
0659f3a0e3 Merge branch 'properly_allocate_unit8_on_cuda' 2015-03-11 10:48:14 -04:00
Robert Maynard
e4bd0132e6 Merge branch 'use_ldg_cuda_load' 2015-03-11 10:47:57 -04:00
Robert Maynard
86dc8f1d38 Move back to thrust::cuda::vector to properly handle allocating uint8's
Our approach of using the underlying allocator inside thrust was a bad approach,
for some reason it fails to properly allocate uint8's or int8's on the correct
boundaries. I expect that this logic is somewhere else in the code and
instead we should use thrust::system::cuda::vector which does this properly.
2015-03-10 10:10:03 -04:00
Robert Maynard
63b1f03187 Simplify the implementation of loading through textures.
We don't need this super complicated system for texture loading.
2015-03-09 16:37:45 -04:00
Robert Maynard
9b49973621 Use __ldg instead of texture object. 2015-03-05 18:31:44 -05:00
Kenneth Moreland
0c49c20035 Merge branch 'fix-copyright-check' 2015-02-11 09:48:39 -07:00
Kenneth Moreland
baf941beb1 Fix the date manipulation on the check copyright script
When a new file is added to VTK-m, the copyright statement should go at
the top of the file. The copyright contains a date. What should that date
be? I usually set the date to the current year so older files will have
an older copyright whereas newer files will have a newer one. The check
copyright script needs to be flexible on the date.

There was an error in the script that was copied over from Dax. It was
checking for the year 2011, the start of the Dax project, and replacing
that in the text. VTK-m started in 2014, so the script really needs to
check for that year instead.
2015-02-11 09:45:36 -07:00
Kenneth Moreland
de34bcbeba Merge branch 'expose-array-handle-allocate' 2015-02-10 16:11:23 -07:00
Kenneth Moreland
6141e83be4 Expose allocation in ArrayHandle.
Add an Allocate method in ArrayHandle that basically forwards the
alllocate request to the storage object. This allows some measure of
control of the array from the control side. You can allocate the array
and set values (by getting the control array portal) if you so desire.
2015-02-10 15:58:41 -07:00
Kenneth Moreland
2f781dfe7a Preserve data in ArrayHandle::ReleaseResourcesExecution
Previously when ReleaseResourcesExecution was called, the method blindly
deleted the execution array regardless of whether there was a valid copy
in the control environment. This could potentially lose data. What if
someone wanted to conserve memory on the device by clearing the array of
an output array?

There is also now an internal method that blindly deletes the array.
This is good for internal functions that are doing something to
invalidate the execution data anyway.
2015-02-10 15:49:55 -07:00
Jeremy Meredith
d807c1d865 map cell is now map topology -- the type of map depends on the passed-in topology. 2015-02-10 15:01:27 -05:00
Jeremy Meredith
6d7e2012ab encapsulating field for data set. 2015-02-10 12:36:10 -05:00
Jeremy Meredith
adf0dfd53e moving cell type into new file. 2015-02-09 17:04:09 -05:00
Kenneth Moreland
68ae208dad Merge branch 'get-portal-control-from-empty-array' 2015-02-09 14:54:50 -07:00
Kenneth Moreland
c224c2b98a Make ArrayHandle work better when uninitialized
Fixed a problem where ArrayHandle would cause a crash if you tried to
get the control portal on an uninitialized array (it was supposed to
throw an exception).

Also changed some methods in ArrayHandle so that they work resonably
without error when used with an uninitialized array. Specifically, the
aforementioned behavior was changed to "allocate" an array of size 0
(that is, call Allocate(0) on the storage object) to return an empty
array portal rather than throw an error. Although this use of
ArrayHandle can be considered erroneous, it is convenient the get an
empty array portal when dealing with a potentially unallocated array
rather than create a special condition.
2015-02-09 14:54:04 -07:00
Jeremy Meredith
f4975e7629 making all map cell related parameters use max cell size in control signature topology template arg. 2015-02-09 16:43:31 -05:00
Jeremy Meredith
5a40bb2657 fit margins. 2015-02-09 15:06:25 -05:00
Jeremy Meredith
27ef6f262f making node value input tuplelen be first argument so that there's no default value. 2015-02-09 15:02:16 -05:00
Jeremy Meredith
fb8639f1eb allow arbitrary length vecs for node value sets. 2015-02-09 14:28:57 -05:00
Jeremy Meredith
c741d3cccf allowing variable number of values to return for ID set. 2015-02-09 13:42:01 -05:00
Jeremy Meredith
c34d64e56c force constructors to array handles to work around stl vector allocator weirdness. 2015-02-05 15:38:49 -05:00
Jeremy Meredith
2caf35602b fixing bug with node values in map cell. 2015-02-05 14:08:37 -05:00
Jeremy Meredith
475c4bb5ee renaming some node references to topology and removing unused files. 2015-02-05 13:45:08 -05:00
Jeremy Meredith
c3ff52353c adding ability to get nodal fields in cell map 2015-02-05 13:41:23 -05:00
Jeremy Meredith
20d9c60398 add direct fetch tag for topology which returns cell type. 2015-02-05 13:30:01 -05:00
Jeremy Meredith
be8e8e0aac adding regular connectivity example 2015-02-05 10:04:39 -05:00
Jeremy Meredith
67557a51bc making connectivity fully general (mixed element meshes) and more ways of getting at it in worklets. 2015-02-04 17:40:31 -05:00
Jeremy Meredith
6bd04efef2 changing mapcell input domain from input cell array to topology GetNumberOfCells. 2015-02-04 15:51:13 -05:00
Jeremy Meredith
2f83e08b7d changing exec output for topology to be Id3. 2015-02-04 15:48:15 -05:00
Jeremy Meredith
6214fa6629 adding execution output value type for node id set. 2015-02-04 15:24:50 -05:00
Jeremy Meredith
da73c67f4a first steps in getting real topology to worklet map cell. 2015-02-04 15:02:59 -05:00
Jeremy Meredith
61852bbb20 back to vtkm::Id as connectivity element type in prep for proper implementation. 2015-02-04 12:41:02 -05:00
Jeremy Meredith
455ef255da forcing connectivity to triangle-only (Id3) to get simple WorkletMapCell. 2015-02-04 11:47:15 -05:00
Jeremy Meredith
5a0a658d7d adding worklet map cell framework (so far, identical to field version). 2015-02-04 11:00:15 -05:00
Dave Pugmire
e3970b09f1 Add a simple worklet that creates a cell-type variable. 2015-02-03 12:02:48 -05:00
Dave Pugmire
d6ccc97304 Add initial data set for unstructured mesh. 2015-02-02 16:44:24 -05:00
Robert Maynard
3e3596ab7a Merge branch 'release_cuda_memory' 2015-01-28 15:55:03 -05:00
Robert Maynard
3c8ce36666 Properly deallocate cuda memory when we are done with it. 2015-01-28 15:54:45 -05:00
Dave Pugmire
1b24b1015f add dataset test file. 2015-01-28 11:50:06 -05:00
Dave Pugmire
09ec291feb fill in the dataset with some values. 2015-01-28 11:46:07 -05:00
Dave Pugmire
01dce61672 data model tweaks.
fix a bad commit.
2015-01-28 11:27:15 -05:00
Dave Pugmire
62f243e819 fix conflicted checkin 2015-01-28 09:28:01 -05:00
Dave Pugmire
cf006085b4 Merge branch 'datamodel-design' of public.kitware.com:vtkm into datamodel-design
Conflicts:
	CMakeLists.txt
2015-01-27 17:02:03 -05:00