In preparation of TBB we need to split the finalize function into reduce
and free. Reduce is used to combine results and free for freeing any
allocated memory.
The reduce function is called to join user data chunk into another, to reduce the
result to the original userdata_chunk memory. These functions should have no side
effects so that they can be run on any thread.
The free functions should free data created during execution (TaskParallelRangeFunc).
Original patch by Brecht van Lommel
{rB61f49db843cf5095203112226ae386f301be1e1a}.
Reviewed By: Brecht van Lommel, Bastien Montagne
Differential Revision: https://developer.blender.org/D7394
Loose vertices and vertices of loose edges callback was not working
correct if some of other callbacks were set to NULL.
Was caused by missing bitmask set in the callbacks which were set
to NULL.
Was affecting quad faces. where 0 was always passed for inner
loops and a wrong one for boundary ones.
In the current code this change shouldn't cause any difference
since the corner index is ignored in the actual callback, but
it is required to have his fixed for an upcoming changes.
Ensure that all threads on a multi-core system are used.
The issue was that BLI_task module was trying to be smart and
used heuristic to find optimal number of iterations per thread.
This heuristic assumes that tasks are light-weight, which is
not a case for subdivision surface.
On a higher subdivision level with a file from T70826 the
evaluation time goes down from 0.25 to 0.17 seconds per modifier
evaluation.
When D6189 is finalized we can being some extra performance
improvement.
We already have different storages for cddata of verts, edges etc.,
'simply' do the same for the mask flags we use all around Blender code
to request some data, or limit some operation to some layers, etc.
Reason we need this is that some cddata types (like Normals) are
actually shared between verts/polys/loops, and we don’t want to generate
clnors everytime we request vnors!
As a side note, this also does final fix to T59338, which was the
trigger for this patch (need to request computed loop normals for
another mesh than evaluated one).
Reviewers: brecht, campbellbarton, sergey
Differential Revision: https://developer.blender.org/D4407
BF-admins agree to remove header information that isn't useful,
to reduce noise.
- BEGIN/END license blocks
Developers should add non license comments as separate comment blocks.
No need for separator text.
- Contributors
This is often invalid, outdated or misleading
especially when splitting files.
It's more useful to git-blame to find out who has developed the code.
See P901 for script to perform these edits.
This switches evaluation of vertices which are on the boundaries
of PTex faces to a single threaded one. While this introduces
some slowdown it fixes ambiguity of PTex index used to evaluate
particular vertex.
Possible alternative solutions would be:
- Do some pre-calculation of index, then do evaluation in threads.
- Try using Gregory patches and see if that makes any affect.
Fix T60235: Flickering of object instances
There is no point having operations that iterate over the whole
bit array as macros, so convert BLI_BITMAP_SET_ALL to a function.
Also, add more utilities for copying and manipulating masks.
Reviewers: brecht, campbellbarton
Differential Revision: https://developer.blender.org/D4101
This is something what we need to know quite often from various places.
Added it as a cached value in Subdiv itself, so it can be queried easily
from any area.
Shouldn't be a problem from memory usage point of view, it's 4MB per
1M faces coarse mesh. This is very low percentage of mesh itself, and
even lower percentage of highres subdivided mesh.
This makes it more generic process to perform actions which
depend on ptex face + (u, v) and on subdivided vertex index.
Currently it is still just a subdivision calculation process,
but same foreach callbacks can easily be used to propagate
displacement from known vertex locations back to displacement
grids.
