As an optimization, dependency graph evaluation skips through
no-op nodes at the scheduling stage. However, that leaves update
flags enabled on the node, and the most problematic one is the
USER_MODIFIED flag, which get flushed to children every time
the no-op node is tagged.
This in turn can cause simulation caches downstream to be
permanently stuck in an outdated state until the depsgraph
is rebuilt and the stuck flag cleared out.
Differential Revision: https://developer.blender.org/D16868
The function is already doing a lot of memory indirections and
sub-optimal lookups, so for the simplicity and robustness of the
system might as well just do copy-on-write update.
This is the conventional way of dealing with unused arguments in C++,
since it works on all compilers.
Regex find and replace: `UNUSED\((\w+)\)` -> `/*$1*/`
The ID nodes will use the provided component name to maintain
the map-based storage, while the component node itself could
override the empty name with a type name.
This lead to situations when it is not possible to lookup
the operation from its owner parameters.
This was essentially a use-after-free issue. When a geometry nodes
group changes it has to be preprocessed again before it can be evaluated.
This part was working, the issue was that parent node groups have to be
preprocessed as well, which was missing. The lazy-function graph cached
on the parent node group was still referencing data that was freed when
the child group changed.
Now the depsgraph makes sure that all relevant geometry node groups are
preprocessed again after a change.
This issue was found by Simon Thommes.
The internal state tracking is not fully suited for such kind
of optimization yet.
It is probably not that much work to make them work, but the
issue caused by the changes is serious enough for the studio
so it feels better to revert changes for now and have a closer
look into remaining issues without pressure.
A regression since ac20970bc2
The issue was caused by depsgraph clearing all id->recalc flags
wrongly assuming that all IDs are fully evaluated.
This change makes it so the depsgraph becomes aware of possibly
incompletely evaluated IDs.
Differential Revision: https://developer.blender.org/D15946
Solves long-standing issue when dependencies of disabled modifiers are
evaluated.
Simple test case: no drivers or animation. Manually enabling modifier
is expected to bring FPS up, enabling modifier will bring FPS (sine
evaluation can not be avoided)
F13336690
More complex test case: modifier visibility is driven by an animated
property. In am ideal world FPS during property being zero is fast
and when property is 1 the FPS is low.
F13336691.
Differential Revision: https://developer.blender.org/D15625
A mistake in the 0dcee6a386 which made specific driven visibility
to work, but did not properly handle actual time-based visibility.
The basic idea of the change is to preserve recalculation flags of
nodes which were tagged for update but were not evaluated due to
visibility constraints. In the file from the report this makes it
so tagging which is done first time ID is in the dependency graph
are handled when the ID actually becomes visible. This is what
solved the root of the problem from the report: there was missing
geometry update since it was "swallowed" by the evaluation during
the object being invisible. In other configurations this change
allows to handle pending geometry updates due to animated modifiers
be handled when object becomes visible without time change.
This change also solves visibility issue of the synchronization
component which also started to be handled badly since the
previous fix attempt. Basically, the needed exception in its
visibility handling did not happen and a regular logic was used
for it.
Tested with files from the T99733, T99976, and from the Heist
project.
Differential Revision: https://developer.blender.org/D15544
The issue was caused by the fact that objects with driven or animated
visibility were considered visible by the dependency graph evaluation.
This change makes it so the dependency graph evaluation is aware of
visibility which might be changing. This is achieved by evaluating the
path of the graph which affects objects visibility and adjusts to it
before evaluating the rest of the graph.
There is some time penalty to this, but there does not seem to be a
way to fully avoid this penalty.
With the production shot from the heist project the FPS drops by a
tenth of a frame (~9.4 vs ~9.3 fps) when adding a driver to an object
which keeps it visible. Note that this is a bit hard to measure since
the FPS fluctuates quite a bit throughout the playback. On the other
hand, having a driver on a visibility of a heavy object from character
and setting visibility to false gives big speedup.
Also worth noting that there is no penalty at all when there are no
animated visibilities in the scene.
Differential Revision: https://developer.blender.org/D15498
Use a shorter/simpler license convention, stops the header taking so
much space.
Follow the SPDX license specification: https://spdx.org/licenses
- C/C++/objc/objc++
- Python
- Shell Scripts
- CMake, GNUmakefile
While most of the source tree has been included
- `./extern/` was left out.
- `./intern/cycles` & `./intern/atomic` are also excluded because they
use different header conventions.
doc/license/SPDX-license-identifiers.txt has been added to list SPDX all
used identifiers.
See P2788 for the script that automated these edits.
Reviewed By: brecht, mont29, sergey
Ref D14069
Part of T91671.
Not much else to say, this is mainly a massive deletion of code.
Note that a few cleanups possible after this proxy removal were kept out
of this commit to try to reduce a bit its size.
Reviewed By: sergey, brecht
Maniphest Tasks: T91671
Differential Revision: https://developer.blender.org/D13995
Blender would have crashed when renaming bone in Edit Mode, Saving, and
than selecting/deselecting.
Caused by a mistake in the 0f89bcdbeb: can not "short-circuit" the
CoW update if it was explicitly requested.
Safest for now solution seems to be to store whether the CoW component
has been explicitly tagged, so that the following configuration can be
supported:
DEG_id_tag_update(id, ID_RECALC_GEOMETRY);
DEG_id_tag_update(id, ID_RECALC_COPY_ON_WRITE);
Differential Revision: https://developer.blender.org/D13966
The evaluated mesh is a result of evaluated modifiers, and referencing
other evaluated IDs such as materials.
It can not be stored in the EditMesh structure which is intended to be
re-used by many areas. Such sharing was causing ownership errors causing
bugs like
T93855: Cycles crash with edit mode and simultaneous viewport and final render
The proposed solution is to store the evaluated edit mesh and its cage in
the object's runtime field. The motivation goes as following:
- It allows to avoid ownership problems like the ones in the linked report.
- Object level is chosen over mesh level is because the evaluated mesh
is affected by modifiers, which are on the object level.
This patch allows to have modifier stack of an object which shares mesh with
an object which is in edit mode to be properly taken into account (before
the change the modifier stack from the active object will be used for all
objects which share the mesh).
There is a change in the way how copy-on-write is handled in the edit mode to
allow proper state update when changing active scene (or having two windows
with different scenes). Previously, the copt-on-write would have been ignored
by skipping tagging CoW component. Now it is ignored from within the CoW
operation callback. This allows to update edit pointers for objects which are
not from the current depsgraph and where the edit_mesh was never assigned in
the case when the depsgraph was evaluated prior the active depsgraph.
There is no user level changes changes expected with the CoW handling changes:
should not affect on neither performance, nor memory consumption.
Tested scenarios:
- Various modifiers configurations of objects sharing mesh and be part of the
same scene.
- Steps from the reports: T93855, T82952, T77359
This also fixes T76609, T72733 and perhaps other reports.
Differential Revision: https://developer.blender.org/D13824
This patch implements the vector types (i.e:`float2`) by making heavy
usage of templating. All vector functions are now outside of the vector
classes (inside the `blender::math` namespace) and are not vector size
dependent for the most part.
In the ongoing effort to make shaders less GL centric, we are aiming
to share more code between GLSL and C++ to avoid code duplication.
####Motivations:
- We are aiming to share UBO and SSBO structures between GLSL and C++.
This means we will use many of the existing vector types and others
we currently don't have (uintX, intX). All these variations were
asking for many more code duplication.
- Deduplicate existing code which is duplicated for each vector size.
- We also want to share small functions. Which means that vector
functions should be static and not in the class namespace.
- Reduce friction to use these types in new projects due to their
incompleteness.
- The current state of the `BLI_(float|double|mpq)(2|3|4).hh` is a
bit of a let down. Most clases are incomplete, out of sync with each
others with different codestyles, and some functions that should be
static are not (i.e: `float3::reflect()`).
####Upsides:
- Still support `.x, .y, .z, .w` for readability.
- Compact, readable and easilly extendable.
- All of the vector functions are available for all the vectors types
and can be restricted to certain types. Also template specialization
let us define exception for special class (like mpq).
- With optimization ON, the compiler unroll the loops and performance
is the same.
####Downsides:
- Might impact debugability. Though I would arge that the bugs are
rarelly caused by the vector class itself (since the operations are
quite trivial) but by the type conversions.
- Might impact compile time. I did not saw a significant impact since
the usage is not really widespread.
- Functions needs to be rewritten to support arbitrary vector length.
For instance, one can't call `len_squared_v3v3` in
`math::length_squared()` and call it a day.
- Type cast does not work with the template version of the `math::`
vector functions. Meaning you need to manually cast `float *` and
`(float *)[3]` to `float3` for the function calls.
i.e: `math::distance_squared(float3(nearest.co), positions[i]);`
- Some parts might loose in readability:
`float3::dot(v1.normalized(), v2.normalized())`
becoming
`math::dot(math::normalize(v1), math::normalize(v2))`
But I propose, when appropriate, to use
`using namespace blender::math;` on function local or file scope to
increase readability.
`dot(normalize(v1), normalize(v2))`
####Consideration:
- Include back `.length()` method. It is quite handy and is more C++
oriented.
- I considered the GLM library as a candidate for replacement. It felt
like too much for what we need and would be difficult to extend / modify
to our needs.
- I used Macros to reduce code in operators declaration and potential
copy paste bugs. This could reduce debugability and could be reverted.
- This touches `delaunay_2d.cc` and the intersection code. I would like
to know @howardt opinion on the matter.
- The `noexcept` on the copy constructor of `mpq(2|3)` is being removed.
But according to @JacquesLucke it is not a real problem for now.
I would like to give a huge thanks to @JacquesLucke who helped during this
and pushed me to reduce the duplication further.
Reviewed By: brecht, sergey, JacquesLucke
Differential Revision: https://developer.blender.org/D13791
This patch implements the vector types (i.e:float2) by making heavy
usage of templating. All vector functions are now outside of the vector
classes (inside the blender::math namespace) and are not vector size
dependent for the most part.
In the ongoing effort to make shaders less GL centric, we are aiming
to share more code between GLSL and C++ to avoid code duplication.
Motivations:
- We are aiming to share UBO and SSBO structures between GLSL and C++.
This means we will use many of the existing vector types and others we
currently don't have (uintX, intX). All these variations were asking
for many more code duplication.
- Deduplicate existing code which is duplicated for each vector size.
- We also want to share small functions. Which means that vector functions
should be static and not in the class namespace.
- Reduce friction to use these types in new projects due to their
incompleteness.
- The current state of the BLI_(float|double|mpq)(2|3|4).hh is a bit of a
let down. Most clases are incomplete, out of sync with each others with
different codestyles, and some functions that should be static are not
(i.e: float3::reflect()).
Upsides:
- Still support .x, .y, .z, .w for readability.
- Compact, readable and easilly extendable.
- All of the vector functions are available for all the vectors types and
can be restricted to certain types. Also template specialization let us
define exception for special class (like mpq).
- With optimization ON, the compiler unroll the loops and performance is
the same.
Downsides:
- Might impact debugability. Though I would arge that the bugs are rarelly
caused by the vector class itself (since the operations are quite trivial)
but by the type conversions.
- Might impact compile time. I did not saw a significant impact since the
usage is not really widespread.
- Functions needs to be rewritten to support arbitrary vector length. For
instance, one can't call len_squared_v3v3 in math::length_squared() and
call it a day.
- Type cast does not work with the template version of the math:: vector
functions. Meaning you need to manually cast float * and (float *)[3] to
float3 for the function calls.
i.e: math::distance_squared(float3(nearest.co), positions[i]);
- Some parts might loose in readability:
float3::dot(v1.normalized(), v2.normalized())
becoming
math::dot(math::normalize(v1), math::normalize(v2))
But I propose, when appropriate, to use
using namespace blender::math; on function local or file scope to
increase readability. dot(normalize(v1), normalize(v2))
Consideration:
- Include back .length() method. It is quite handy and is more C++
oriented.
- I considered the GLM library as a candidate for replacement.
It felt like too much for what we need and would be difficult to
extend / modify to our needs.
- I used Macros to reduce code in operators declaration and potential
copy paste bugs. This could reduce debugability and could be reverted.
- This touches delaunay_2d.cc and the intersection code. I would like to
know @Howard Trickey (howardt) opinion on the matter.
- The noexcept on the copy constructor of mpq(2|3) is being removed.
But according to @Jacques Lucke (JacquesLucke) it is not a real problem
for now.
I would like to give a huge thanks to @Jacques Lucke (JacquesLucke) who
helped during this and pushed me to reduce the duplication further.
Reviewed By: brecht, sergey, JacquesLucke
Differential Revision: http://developer.blender.org/D13791
Continuation of the D13404 which finished the design of not having
geometry-level nodes dependent on object-level.
Differential Revision: https://developer.blender.org/D13405
Goals of this refactor:
* More unified approach to updating everything that needs to be updated
after a change in a node tree.
* The updates should happen in the correct order and quadratic or worse
algorithms should be avoided.
* Improve detection of changes to the output to avoid tagging the depsgraph
when it's not necessary.
* Move towards a more declarative style of defining nodes by having a
more centralized update procedure.
The refactor consists of two main parts:
* Node tree tagging and update refactor.
* Generally, when changes are done to a node tree, it is tagged dirty
until a global update function is called that updates everything in
the correct order.
* The tagging is more fine-grained compared to before, to allow for more
precise depsgraph update tagging.
* Depsgraph changes.
* The shading specific depsgraph node for node trees as been removed.
* Instead, there is a new `NTREE_OUTPUT` depsgrap node, which is only
tagged when the output of the node tree changed (e.g. the Group Output
or Material Output node).
* The copy-on-write relation from node trees to the data block they are
embedded in is now non-flushing. This avoids e.g. triggering a material
update after the shader node tree changed in unrelated ways. Instead
the material has a flushing relation to the new `NTREE_OUTPUT` node now.
* The depsgraph no longer reports data block changes through to cycles
through `Depsgraph.updates` when only the node tree changed in ways
that do not affect the output.
Avoiding unnecessary updates seems to work well for geometry nodes and cycles.
The situation is a bit worse when there are drivers on the node tree, but that
could potentially be improved separately in the future.
Avoiding updates in eevee and the compositor is more tricky, but also less urgent.
* Eevee updates are triggered by calling `DRW_notify_view_update` in
`ED_render_view3d_update` indirectly from `DEG_editors_update`.
* Compositor updates are triggered by `ED_node_composite_job` in `node_area_refresh`.
This is triggered by calling `ED_area_tag_refresh` in `node_area_listener`.
Removing updates always has the risk of breaking some dependency that no
one was aware of. It's not unlikely that this will happen here as well. Adding
back missing updates should be quite a bit easier than getting rid of
unnecessary updates though.
Differential Revision: https://developer.blender.org/D13246
- Added space below non doc-string comments to make it clear
these aren't comments for the symbols directly below them.
- Use doxy sections for some headers.
Ref T92709
The are few things in the dependency graph which lead to the issue:
- IDs are only built once.
- Object-data level (Armature, i,e,) builder dependent on the object
visibility.
This caused issues when an armature is first built as not directly
visible (via driver, i.e.) and then was built as a directly visible.
This did not update visibility flag on the node for the custom shape
object.
The idea behind the fix is to go away form passing object visibility
flag to the geometry-level builders and instead rely on the common
visibility flush post-processing to make sure certain objects are
fully visible when needed.
This is the safest minimal part of the change for 3.0 release which
acts as an additional way to ensure visibility. This means that it
might not be a complete fix (if some configuration was overseen) but
it should not make currently working cases to not work.
The fix should also make modifiers used on rigify widgets to work.
The more complete fix will have `is_object_visible` argument removed
from the geometry-level builder functions.
Differential Revision: https://developer.blender.org/D13404
During a mesh transformation in edit mode (Move, Rotate...), only part of
the batch cache needs to be updated.
This commit allows only update only the drawn batches seen in
`BKE_object_data_eval_batch_cache_deform_tag` if the new
`ID_RECALC_GEOMETRY_DEFORM` flag is used.
This new flag is used in the transforms operation for edit-mesh and
results in 1.6x overall speedup in heavy subdiv cube.
Differential Revision: https://developer.blender.org/D11599
This resolves a bottleneck where every update while transforming
copied the entire mesh data-block, which isn't needed as the edit-mesh
is the source of the data being edited.
Testing shows a significant overall speedup when transforming:
- ~1.5x with a subdivided cube 1.5 million vertices.
- ~3.0x with the spring mesh (edit-mode with modifiers disabled,
duplicated 10x to drop performance).
Reviewed By: sergey
Ref D11337
Avoid computationally expensive copying operations
when only some settings have been modified.
This is done by adding support for updating parameters
without tagging for copy-on-write.
Currently only mesh data blocks are supported,
other data-blocks can be added individually.
This prepares for changing values such as edit-mesh auto-smooth angle
in edit-mode without duplicating all mesh-data.
The benefit will only be seen when the user interface no longer tags
all ID's for copy on write updates.
ID_RECALC_GEOMETRY_ALL_MODES has been added to support situations
where non edit-mode geometry is modified in edit-mode.
While this isn't something user are likely to do,
Python scripts may change the underlying mesh.
Reviewed By: sergey
Ref D11377
