Attribute Math: Improve performance of mixer in some cases

The `DefaultMixer` for mixing generic data types has some issues:
1. The full buffer is always zeroed, even if only some is used.
2. Finalizing also works on all values, even if only some are used.
3. "mixing" doesn't allow setting the first value, requiring that
everything is cleared beforehand.

This commit adds the following functionality:
1. Constructor with the specified `IndexMask` for preliminary zeroing.
2. `set` method to overwrite the value.
3. `finalize` with the specified mask to process a subset of values.

This is useful in situations where you want to use the
DefaultMixer without having to overwrite all the values many times.

A performance improvement was observed for NURBS curve evaluation and
attribute interpolation from the point to curve domain of about 15% and
35% respectively (100,000 curves).

Differential Revision: https://developer.blender.org/D15434

1 files changed, 21 insertions, 19 deletions

diff --git a/source/blender/blenkernel/intern/curve_nurbs.cc b/source/blender/blenkernel/intern/curve_nurbs.cc
index 3ab6fb01ea5..62d5682da0f 100644
--- a/source/blender/blenkernel/intern/curve_nurbs.cc
+++ b/source/blender/blenkernel/intern/curve_nurbs.cc
@@ -4,8 +4,9 @@
  * \ingroup bke
  */
 
-#include "BKE_attribute_math.hh"
+#include "BLI_task.hh"
 
+#include "BKE_attribute_math.hh"
 #include "BKE_curves.hh"
 
 namespace blender::bke::curves::nurbs {
@@ -192,16 +193,16 @@ static void interpolate_to_evaluated(const BasisCache &basis_cache,
 {
   attribute_math::DefaultMixer<T> mixer{dst};
 
-  for (const int i : dst.index_range()) {
-    Span<float> point_weights = basis_cache.weights.as_span().slice(i * order, order);
-
-    for (const int j : point_weights.index_range()) {
-      const int point_index = (basis_cache.start_indices[i] + j) % src.size();
-      mixer.mix_in(i, src[point_index], point_weights[j]);
+  threading::parallel_for(dst.index_range(), 128, [&](const IndexRange range) {
+    for (const int i : range) {
+      Span<float> point_weights = basis_cache.weights.as_span().slice(i * order, order);
+      for (const int j : point_weights.index_range()) {
+        const int point_index = (basis_cache.start_indices[i] + j) % src.size();
+        mixer.mix_in(i, src[point_index], point_weights[j]);
+      }
     }
-  }
-
-  mixer.finalize();
+    mixer.finalize(range);
+  });
 }
 
 template<typename T>
@@ -213,17 +214,18 @@ static void interpolate_to_evaluated_rational(const BasisCache &basis_cache,
 {
   attribute_math::DefaultMixer<T> mixer{dst};
 
-  for (const int i : dst.index_range()) {
-    Span<float> point_weights = basis_cache.weights.as_span().slice(i * order, order);
+  threading::parallel_for(dst.index_range(), 128, [&](const IndexRange range) {
+    for (const int i : range) {
+      Span<float> point_weights = basis_cache.weights.as_span().slice(i * order, order);
 
-    for (const int j : point_weights.index_range()) {
-      const int point_index = (basis_cache.start_indices[i] + j) % src.size();
-      const float weight = point_weights[j] * control_weights[point_index];
-      mixer.mix_in(i, src[point_index], weight);
+      for (const int j : point_weights.index_range()) {
+        const int point_index = (basis_cache.start_indices[i] + j) % src.size();
+        const float weight = point_weights[j] * control_weights[point_index];
+        mixer.mix_in(i, src[point_index], weight);
+      }
     }
-  }
-
-  mixer.finalize();
+    mixer.finalize(range);
+  });
 }
 
 void interpolate_to_evaluated(const BasisCache &basis_cache,