archive/blender-archive
Archived
1
1
Fork 0
Code Pull Requests Packages Activity
This repository has been archived on 2023-10-09. You can view files and clone it, but cannot push or open issues or pull requests.
Files
905d1c71b9f3b0543357c0ea190d77faca40d54d
blender-archive/source/blender/functions/FN_lazy_function_execute.hh

130 lines
4.8 KiB
C++
Raw Normal View History

Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
/* SPDX-License-Identifier: GPL-2.0-or-later */
#pragma once
/** \file
* \ingroup fn
*
* This file contains common utilities for actually executing a lazy-function.
*/
#include "BLI_parameter_pack_utils.hh"
#include "FN_lazy_function.hh"
namespace blender::fn::lazy_function {
/**
* Most basic implementation of #Params. It does not actually implement any logic for how to
* retrieve inputs or set outputs. Instead, code using #BasicParams has to implement that.
*/
class BasicParams : public Params {
private:
const Span<GMutablePointer> inputs_;
const Span<GMutablePointer> outputs_;
MutableSpan<std::optional<ValueUsage>> input_usages_;
Span<ValueUsage> output_usages_;
MutableSpan<bool> set_outputs_;
public:
BasicParams(const LazyFunction &fn,
const Span<GMutablePointer> inputs,
const Span<GMutablePointer> outputs,
MutableSpan<std::optional<ValueUsage>> input_usages,
Span<ValueUsage> output_usages,
MutableSpan<bool> set_outputs);
void *try_get_input_data_ptr_impl(const int index) const override;
void *try_get_input_data_ptr_or_request_impl(const int index) override;
void *get_output_data_ptr_impl(const int index) override;
void output_set_impl(const int index) override;
bool output_was_set_impl(const int index) const override;
ValueUsage get_output_usage_impl(const int index) const override;
void set_input_unused_impl(const int index) override;
Geometry Nodes: improve evaluator with lazy threading In large node setup the threading overhead was sometimes very significant. That's especially true when most nodes do very little work. This commit improves the scheduling by not using multi-threading in many cases unless it's likely that it will be worth it. For more details see the comments in `BLI_lazy_threading.hh`. Differential Revision: https://developer.blender.org/D15976
2022-09-20 10:59:12 +02:00
bool try_enable_multi_threading_impl() override;
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
};
namespace detail {
/**
* Utility to implement #execute_lazy_function_eagerly.
*/
template<typename... Inputs, typename... Outputs, size_t... InIndices, size_t... OutIndices>
inline void execute_lazy_function_eagerly_impl(
const LazyFunction &fn,
UserData *user_data,
Functions: improve handling of thread-local data in lazy functions The main goal here is to reduce the number of times thread-local data has to be looked up using e.g. `EnumerableThreadSpecific.local()`. While this isn't a bottleneck in many cases, it is when the action performed on the local data is very short and that happens very often (e.g. logging used sockets during geometry nodes evaluation). The solution is to simply pass the thread-local data as parameter to many functions that use it, instead of looking it up in those functions which generally is more costly. The lazy-function graph executor now only looks up the local data if it knows that it might be on a new thread, otherwise it uses the local data retrieved earlier. Alongside with `UserData` there is `LocalUserData` now. This allows users of the lazy-function evaluation (such as geometry nodes) to have custom thread-local data that is passed to all the lazy-functions automatically. This is used for logging now.
2023-05-09 13:10:47 +02:00
LocalUserData *local_user_data,
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
std::tuple<Inputs...> &inputs,
std::tuple<Outputs *...> &outputs,
std::index_sequence<InIndices...> /* in_indices */,
std::index_sequence<OutIndices...> /* out_indices */)
{
constexpr size_t InputsNum = sizeof...(Inputs);
constexpr size_t OutputsNum = sizeof...(Outputs);
std::array<GMutablePointer, InputsNum> input_pointers;
std::array<GMutablePointer, OutputsNum> output_pointers;
std::array<std::optional<ValueUsage>, InputsNum> input_usages;
std::array<ValueUsage, OutputsNum> output_usages;
std::array<bool, OutputsNum> set_outputs;
(
[&]() {
constexpr size_t I = InIndices;
Cleanup: add hint that typedef is used as compiler bug workaround https://developercommunity.visualstudio.com/t/Alias-template-inside-fold-expression-fa/10040507
2022-09-20 10:42:25 +02:00
/* Use `typedef` instead of `using` to work around a compiler bug. */
Workaround for msvc compiler bug This is the same issue as in rB2e8089b6bf50.
2022-09-15 09:33:57 +02:00
typedef Inputs T;
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
const CPPType &type = CPPType::get<T>();
input_pointers[I] = {type, &std::get<I>(inputs)};
}(),
...);
(
[&]() {
constexpr size_t I = OutIndices;
Cleanup: add hint that typedef is used as compiler bug workaround https://developercommunity.visualstudio.com/t/Alias-template-inside-fold-expression-fa/10040507
2022-09-20 10:42:25 +02:00
/* Use `typedef` instead of `using` to work around a compiler bug. */
Workaround for msvc compiler bug This is the same issue as in rB2e8089b6bf50.
2022-09-15 09:33:57 +02:00
typedef Outputs T;
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
const CPPType &type = CPPType::get<T>();
output_pointers[I] = {type, std::get<I>(outputs)};
}(),
...);
output_usages.fill(ValueUsage::Used);
set_outputs.fill(false);
LinearAllocator<> allocator;
Functions: improve handling of thread-local data in lazy functions The main goal here is to reduce the number of times thread-local data has to be looked up using e.g. `EnumerableThreadSpecific.local()`. While this isn't a bottleneck in many cases, it is when the action performed on the local data is very short and that happens very often (e.g. logging used sockets during geometry nodes evaluation). The solution is to simply pass the thread-local data as parameter to many functions that use it, instead of looking it up in those functions which generally is more costly. The lazy-function graph executor now only looks up the local data if it knows that it might be on a new thread, otherwise it uses the local data retrieved earlier. Alongside with `UserData` there is `LocalUserData` now. This allows users of the lazy-function evaluation (such as geometry nodes) to have custom thread-local data that is passed to all the lazy-functions automatically. This is used for logging now.
2023-05-09 13:10:47 +02:00
Context context(fn.init_storage(allocator), user_data, local_user_data);
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
BasicParams params{
fn, input_pointers, output_pointers, input_usages, output_usages, set_outputs};
fn.execute(params, context);
fn.destruct_storage(context.storage);
Functions: support cycles in lazy-function graph Lazy-function graphs are now evaluated properly even if they contain cycles. Note that cycles are only ok if there is no data dependency cycle. For example, a node might output something that is fed back into itself. As long as the output can be computed without the input that it feeds into, everything is ok. The code that builds the graph is responsible for making sure that there are no actual data dependencies.
2022-12-29 16:38:18 +01:00
Cleanup: comment blocks
2023-03-09 10:39:49 +11:00
/* Make sure all outputs have been computed. */
Fix build error in debug mode due to wrong code in assert
2023-01-04 10:43:31 +01:00
BLI_assert(!Span<bool>(set_outputs).contains(false));
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
}
} // namespace detail
/**
* In some cases (mainly for tests), the set of inputs and outputs for a lazy-function is known at
* compile time and one just wants to compute the outputs based on the inputs, without any
Cleanup: spelling in comments
2022-09-13 18:00:44 +10:00
* laziness.
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
*
* This function does exactly that. It takes all inputs in a tuple and writes the outputs to points
* provided in a second tuple. Since all inputs have to be provided, the lazy-function has to
* compute all outputs.
*/
template<typename... Inputs, typename... Outputs>
inline void execute_lazy_function_eagerly(const LazyFunction &fn,
UserData *user_data,
Functions: improve handling of thread-local data in lazy functions The main goal here is to reduce the number of times thread-local data has to be looked up using e.g. `EnumerableThreadSpecific.local()`. While this isn't a bottleneck in many cases, it is when the action performed on the local data is very short and that happens very often (e.g. logging used sockets during geometry nodes evaluation). The solution is to simply pass the thread-local data as parameter to many functions that use it, instead of looking it up in those functions which generally is more costly. The lazy-function graph executor now only looks up the local data if it knows that it might be on a new thread, otherwise it uses the local data retrieved earlier. Alongside with `UserData` there is `LocalUserData` now. This allows users of the lazy-function evaluation (such as geometry nodes) to have custom thread-local data that is passed to all the lazy-functions automatically. This is used for logging now.
2023-05-09 13:10:47 +02:00
LocalUserData *local_user_data,
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
std::tuple<Inputs...> inputs,
std::tuple<Outputs *...> outputs)
{
BLI_assert(fn.inputs().size() == sizeof...(Inputs));
BLI_assert(fn.outputs().size() == sizeof...(Outputs));
detail::execute_lazy_function_eagerly_impl(fn,
user_data,
Functions: improve handling of thread-local data in lazy functions The main goal here is to reduce the number of times thread-local data has to be looked up using e.g. `EnumerableThreadSpecific.local()`. While this isn't a bottleneck in many cases, it is when the action performed on the local data is very short and that happens very often (e.g. logging used sockets during geometry nodes evaluation). The solution is to simply pass the thread-local data as parameter to many functions that use it, instead of looking it up in those functions which generally is more costly. The lazy-function graph executor now only looks up the local data if it knows that it might be on a new thread, otherwise it uses the local data retrieved earlier. Alongside with `UserData` there is `LocalUserData` now. This allows users of the lazy-function evaluation (such as geometry nodes) to have custom thread-local data that is passed to all the lazy-functions automatically. This is used for logging now.
2023-05-09 13:10:47 +02:00
local_user_data,
Geometry Nodes: new evaluation system This refactors the geometry nodes evaluation system. No changes for the user are expected. At a high level the goals are: * Support using geometry nodes outside of the geometry nodes modifier. * Support using the evaluator infrastructure for other purposes like field evaluation. * Support more nodes, especially when many of them are disabled behind switch nodes. * Support doing preprocessing on node groups. For more details see T98492. There are fairly detailed comments in the code, but here is a high level overview for how it works now: * There is a new "lazy-function" system. It is similar in spirit to the multi-function system but with different goals. Instead of optimizing throughput for highly parallelizable work, this system is designed to compute only the data that is actually necessary. What data is necessary can be determined dynamically during evaluation. Many lazy-functions can be composed in a graph to form a new lazy-function, which can again be used in a graph etc. * Each geometry node group is converted into a lazy-function graph prior to evaluation. To evaluate geometry nodes, one then just has to evaluate that graph. Node groups are no longer inlined into their parents. Next steps for the evaluation system is to reduce the use of threads in some situations to avoid overhead. Many small node groups don't benefit from multi-threading at all. This is much easier to do now because not everything has to be inlined in one huge node tree anymore. Differential Revision: https://developer.blender.org/D15914
2022-09-13 08:44:26 +02:00
inputs,
outputs,
std::make_index_sequence<sizeof...(Inputs)>(),
std::make_index_sequence<sizeof...(Outputs)>());
}
} // namespace blender::fn::lazy_function
Copy Permalink