Age | Commit message (Collapse) | Author |
|
My benchmark which spend most time preparing function parameters
takes `250 ms` now, from `510 ms` before. This is mainly achieved by
doing less unnecessary work and by giving the compiler more inlined
code to optimize.
* Reserve correct vector sizes and use unchecked `append` function.
* Construct `GVArray` parameters directly in the vector, instead of
moving/copying them in the vector afterwards.
* Inline some constructors, because that allows the compiler understand
what is happening, resulting in less code.
This probably has negilible impact on the user experience currently,
because there are other bottlenecks.
Differential Revision: https://developer.blender.org/D15009
|
|
Goals:
* Better high level control over where devirtualization occurs. There is always
a trade-off between performance and compile-time/binary-size.
* Simplify using array devirtualization.
* Better performance for cases where devirtualization wasn't used before.
Many geometry nodes accept fields as inputs. Internally, that means that the
execution functions have to accept so called "virtual arrays" as inputs. Those
can be e.g. actual arrays, just single values, or lazily computed arrays.
Due to these different possible virtual arrays implementations, access to
individual elements is slower than it would be if everything was just a normal
array (access does through a virtual function call). For more complex execution
functions, this overhead does not matter, but for small functions (like a simple
addition) it very much does. The virtual function call also prevents the compiler
from doing some optimizations (e.g. loop unrolling and inserting simd instructions).
The solution is to "devirtualize" the virtual arrays for small functions where the
overhead is measurable. Essentially, the function is generated many times with
different array types as input. Then there is a run-time dispatch that calls the
best implementation. We have been doing devirtualization in e.g. math nodes
for a long time already. This patch just generalizes the concept and makes it
easier to control. It also makes it easier to investigate the different trade-offs
when it comes to devirtualization.
Nodes that we've optimized using devirtualization before didn't get a speedup.
However, a couple of nodes are using devirtualization now, that didn't before.
Those got a 2-4x speedup in common cases.
* Map Range
* Random Value
* Switch
* Combine XYZ
Differential Revision: https://developer.blender.org/D14628
|
|
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
|
|
The idea behind this change is the same as in
rB6ee2abde82ef121cd6e927995053ac33afdbb438.
A `MultiFunction::debug_parameter_name` method could be
added separately when necessary.
|
|
Previously, the function names were stored in `std::string` and were often
created dynamically (especially when the function just output a constant).
This resulted in a lot of overhead.
Now the function name is just a `const char *` that should be statically
allocated. This is good enough for the majority of cases. If a multi-function
needs a more dynamic name, it can override the `MultiFunction::debug_name`
method.
In my test file with >400,000 simple math nodes, the execution time improves from
3s to 1s.
|
|
The multi-function network system was able to compose multiple
multi-functions into a new one and to evaluate that efficiently.
This functionality was heavily used by the particle nodes prototype
a year ago. However, since then we only used multi-functions
without the need to compose them in geometry nodes.
The upcoming "fields" in geometry nodes will need a way to
compose multi-functions again. Unfortunately, the code removed
in this commit was not ideal for this different kind of function
composition. I've been working on an alternative that will be added
separately when it becomes needed.
I've had to update all the function nodes, because their interface
depended on the multi-function network data structure a bit.
The actual multi-function implementations are still the same though.
|
|
Previously, the signature of a `MultiFunction` was always embedded into the function.
There are two issues with that. First, `MFSignature` is relatively large, because it contains
multiple strings and vectors. Secondly, constructing it can add overhead that should not
be necessary, because often the same signature can be reused.
The solution is to only keep a pointer to a signature in `MultiFunction` that is set during
construction. Child classes are responsible for making sure that the signature lives
long enough. In most cases, the signature is either embedded into the child class or
it is allocated statically (and is only created once).
|
|
When a function is executed for many elements (e.g. per point) it is often the case
that some parameters are different for every element and other parameters are
the same (there are some more less common cases). To simplify writing such
functions one can use a "virtual array". This is a data structure that has a value
for every index, but might not be stored as an actual array internally. Instead, it
might be just a single value or is computed on the fly. There are various tradeoffs
involved when using this data structure which are mentioned in `BLI_virtual_array.hh`.
It is called "virtual", because it uses inheritance and virtual methods.
Furthermore, there is a new virtual vector array data structure, which is an array
of vectors. Both these types have corresponding generic variants, which can be used
when the data type is not known at compile time. This is typically the case when
building a somewhat generic execution system. The function system used these virtual
data structures before, but now they are more versatile.
I've done this refactor in preparation for the attribute processor and other features of
geometry nodes. I moved the typed virtual arrays to blenlib, so that they can be used
independent of the function system.
One open question for me is whether all the generic data structures (and `CPPType`)
should be moved to blenlib as well. They are well isolated and don't really contain
any business logic. That can be done later if necessary.
|
|
|
|
This replaces header include guards with `#pragma once`.
A couple of include guards are not removed yet (e.g. `__RNA_TYPES_H__`),
because they are used in other places.
This patch has been generated by P1561 followed by `make format`.
Differential Revision: https://developer.blender.org/D8466
|
|
|
|
|
|
Object sockets work now, but only the new Object Transforms and the
Particle Mesh Emitter node use it. The emitter does not actually
use the mesh surface yet. Instead, new particles are just emitted around
the origin of the object.
Internally, handles to object data blocks are passed around in the network,
instead of raw object pointers. Using handles has a couple of benefits:
* The caller of the function has control over which handles can be resolved
and therefore limit access to specific data. The set of data blocks that
is accessed by a node tree should be known statically. This is necessary
for a proper integration with the dependency graph.
* When the pointer to an object changes (e.g. after restarting Blender),
all handles are still valid.
* When an object is deleted, the handle is invalidated without causing crashes.
* The handle is just an integer that can be stored per particle and can be cached easily.
The mapping between handles and their corresponding data blocks is
stored in the Simulation data block.
|
|
Those are useful when you have to create containers with static
storage duration. If those would use Blender's guarded allocator,
it would report memory leaks, that are not actually leaks.
|
|
This updates the usage of integer types in code I wrote according to our new style guides.
Major changes:
* Use signed instead of unsigned integers in many places.
* C++ containers in blenlib use `int64_t` for size and indices now (instead of `uint`).
* Hash values for C++ containers are 64 bit wide now (instead of 32 bit).
I do hope that I broke no builds, but it is quite likely that some compiler reports
slightly different errors. Please let me know when there are any errors. If the fix
is small, feel free to commit it yourself.
I compiled successfully on linux with gcc and on windows.
|
|
|
|
|
|
This allows multi-functions to have static storage duration.
|
|
Most of this code is covered by unit tests.
|
|
This adds the `MultiFunction` type and some smallish utility types that it uses.
A `MultiFunction` encapsulates a function that is optimized for throughput by
always processing many elements at once.
This is an important part of the new particle system, because it allows us to
execute user generated node trees for many particles efficiently.
Reviewers: brecht
Differential Revision: https://developer.blender.org/D8030
|