Age | Commit message (Collapse) | Author |
|
|
|
This hopefully fixes T48383 by avoiding two numerical problems that I found in the volume code.
Reviewers: sergey, dingto, brecht
Reviewed By: sergey, dingto, brecht
Maniphest Tasks: T48383
Differential Revision: https://developer.blender.org/D2051
|
|
OpenCL seems to work fine here, and for some reason that comparison was
giving compilation error on OpenCL here.
Better to compile OpenCL kernel than to be fully robust to weird corner
cases.
|
|
A few places still needed to be updated to use the new Mesh::num_triangles()
method; wrong number from triangles.size() was causing crashes.
|
|
The original quad intersection test works by just testing against the two triangles that define the quad.
However, in this case it's actually faster to use the same test that's also used for portals: Determining
the distance to the plane in which the quad lies, calculating the hitpoint and checking whether it's in the
quad by projecting onto the sides.
Reviewers: brecht, sergey, dingto
Reviewed By: dingto
Differential Revision: https://developer.blender.org/D2045
|
|
Was some accumulated precision error happening.
|
|
|
|
|
|
In certain types of animation it's possible to have some objects
scaling to zero. In this case we can save render times by avoid
traversing such instances.
Better to do ti ahead of a time, so traversal stays simple.
Reviewers: lukasstockner97, dingto, brecht
Reviewed By: brecht
Differential Revision: https://developer.blender.org/D2048
|
|
Currently for windows only, this is an initial commit towards native
support of NUMA.
Current commit makes it so Cycles will use all logical processors on
Windows running on system with more than 64 threads.
Reviewers: juicyfruit, dingto, lukasstockner97, maiself, brecht
Subscribers: LazyDodo
Differential Revision: https://developer.blender.org/D2049
|
|
not used.
|
|
|
|
the standard cmake CUDA/NVCC arguments flag allowing 2015 build to use
msvc 2013 for cuda
|
|
It is really handy to know which one is display when having two cards of
same type in the machine.
|
|
Still not sure how to properly solve the issue, needs some trickery to get
actual optimized values from intersection function (using printf() avoids
some optimization and makes stuff render correct).
For the time being let's just simplify check.
|
|
|
|
|
|
- add_vertex() can be called from split_vertex() which does not guarantee
to have properly pre-allocate arrays.
- Need to check whether Cycles is compiled with OSL in XML reader.
|
|
Some of these values can get quite large and are hard to read, adding this
makes it easy to read them at a glance.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D2039
|
|
We had transparent shadows disabled for some time because they were causing
drivers to crash. Can't reproduce that issue anymore with current drivers,
so will enable them and see how it goes.
|
|
|
|
It was possible to have issues in cases when several child dependencies
goes to IDs with different layers. In this case order of flushing was not
really well defined, which could lead to cases when indirect dependency
via invisible object wouldn't work.
Need some sort of barrier to prevent scheduling of parent nodes for until
all children are done, but that's becoming quite nasty thing to implement.
Added a temp field to component for now. maybe it's not so crazy actually
and we might use it for evaluation as well, so we wouldn't flush updates
to components which does not affect visible stuff.
|
|
|
|
|
|
types.
|
|
|
|
|
|
|
|
Differential Revision: https://developer.blender.org/D2016
|
|
separate some arrays.
Differential Revision: https://developer.blender.org/D2016
|
|
Differential Revision: https://developer.blender.org/D2016
|
|
Differential Revision: https://developer.blender.org/D2016
|
|
This reverts 72592cfb, needs more refactoring to do that cleanup.
|
|
Now I can start adding half float without adding even bigger mess to all these functions. ;)
|
|
|
|
Also make prefix consistent, so it starts with either TEX_NUM or TEX_START, followed by texture type and architecture.
|
|
New device! Lots of buttons!
|
|
|
|
|
|
|
|
Saves about 15% for the branched path kernel.
|
|
|
|
|
|
|
|
|
|
Volumes might be small, makes sense to use more precision here.
|
|
Based on patch from Lawrence D'Oliveiro (ldo) in T48499.
|
|
57% less for path and 48% less for branched path.
|
|
Some closures were missing from calculation, leading to an array
under-allocation, presumable causing memory corruption issues with
emission shaders on OpenCL and was causing issues with Volume 3D
textures with CUDA.
The issue was identified by Thomas Dinges, the patch is different
from the original D2006. See the brief discussion there. Current
approach is similar (or the same) as Brecht suggested.
|
|
In Cycle standalone, the xml import of uv's is done in a way that
a vertex could only have one unique uv coordinate. In practice it
is most of the time not the case.
Patch by Laurent Boiron, thanks!
|