Sculpt: use TBB instead of BLI_task for multithreading

This solves performance issues on some computers where there is significant
threading overhead. Rather than doing the complicated work of optimizing our
own task scheduler, use TBB which appears to work well. The downside is that
we have another thread pool, but it is already there when using OpenVDB voxel
remesh.

For future releases we can switch to using TBB to replace our task scheduler
implementation entirely, and use the same thread pool for BLI_task, Cycles,
Mantaflow, etc.

Differential Revision: https://developer.blender.org/D6030

9 files changed, 385 insertions, 239 deletions

diff --git a/source/blender/blenkernel/BKE_pbvh.h b/source/blender/blenkernel/BKE_pbvh.h
index 042b78cd06e..13adb868c01 100644
--- a/source/blender/blenkernel/BKE_pbvh.h
+++ b/source/blender/blenkernel/BKE_pbvh.h
@@ -28,6 +28,10 @@
 /* For embedding CCGKey in iterator. */
 #include "BKE_ccg.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 struct BMLog;
 struct BMesh;
 struct CCGElem;
@@ -45,6 +49,7 @@ struct PBVH;
 struct PBVHNode;
 struct SubdivCCG;
 struct TaskParallelSettings;
+struct TaskParallelTLS;
 
 typedef struct PBVH PBVH;
 typedef struct PBVHNode PBVHNode;
@@ -432,14 +437,39 @@ void BKE_pbvh_node_get_bm_orco_data(PBVHNode *node,
 
 bool BKE_pbvh_node_vert_update_check_any(PBVH *bvh, PBVHNode *node);
 
-void BKE_pbvh_parallel_range_settings(struct TaskParallelSettings *settings,
-                                      bool use_threading,
-                                      int totnode);
-
 // void BKE_pbvh_node_BB_reset(PBVHNode *node);
 // void BKE_pbvh_node_BB_expand(PBVHNode *node, float co[3]);
 
 bool pbvh_has_mask(PBVH *bvh);
 void pbvh_show_mask_set(PBVH *bvh, bool show_mask);
 
+/* Parallelization */
+typedef void (*PBVHParallelRangeFunc)(void *__restrict userdata,
+                                      const int iter,
+                                      const struct TaskParallelTLS *__restrict tls);
+typedef void (*PBVHParallelReduceFunc)(const void *__restrict userdata,
+                                       void *__restrict chunk_join,
+                                       void *__restrict chunk);
+
+typedef struct PBVHParallelSettings {
+  bool use_threading;
+  void *userdata_chunk;
+  size_t userdata_chunk_size;
+  PBVHParallelReduceFunc func_reduce;
+} PBVHParallelSettings;
+
+void BKE_pbvh_parallel_range_settings(struct PBVHParallelSettings *settings,
+                                      bool use_threading,
+                                      int totnode);
+
+void BKE_pbvh_parallel_range(const int start,
+                             const int stop,
+                             void *userdata,
+                             PBVHParallelRangeFunc func,
+                             const struct PBVHParallelSettings *settings);
+
+#ifdef __cplusplus
+}
+#endif
+
 #endif /* __BKE_PBVH_H__ */
diff --git a/source/blender/blenkernel/CMakeLists.txt b/source/blender/blenkernel/CMakeLists.txt
index 47b44c3828a..ec4246f5dba 100644
--- a/source/blender/blenkernel/CMakeLists.txt
+++ b/source/blender/blenkernel/CMakeLists.txt
@@ -183,6 +183,7 @@ set(SRC
   intern/particle_system.c
   intern/pbvh.c
   intern/pbvh_bmesh.c
+  intern/pbvh_parallel.cc
   intern/pointcache.c
   intern/report.c
   intern/rigidbody.c
@@ -637,6 +638,14 @@ if(WITH_QUADRIFLOW)
   add_definitions(-DWITH_QUADRIFLOW)
 endif()
 
+if(WITH_TBB)
+  add_definitions(-DWITH_TBB)
+
+  list(APPEND INC_SYS
+    ${TBB_INCLUDE_DIRS}
+  )
+endif()
+
 ## Warnings as errors, this is too strict!
 #if(MSVC)
 #   set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} /WX")
diff --git a/source/blender/blenkernel/intern/pbvh.c b/source/blender/blenkernel/intern/pbvh.c
index 4e193e35109..01612ded396 100644
--- a/source/blender/blenkernel/intern/pbvh.c
+++ b/source/blender/blenkernel/intern/pbvh.c
@@ -1096,12 +1096,11 @@ static void pbvh_faces_update_normals(PBVH *bvh, PBVHNode **nodes, int totnode)
       .vnors = vnors,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, true, totnode);
 
-  BLI_task_parallel_range(0, totnode, &data, pbvh_update_normals_accum_task_cb, &settings);
-
-  BLI_task_parallel_range(0, totnode, &data, pbvh_update_normals_store_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, pbvh_update_normals_accum_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, pbvh_update_normals_store_task_cb, &settings);
 
   MEM_freeN(vnors);
 }
@@ -1151,9 +1150,9 @@ static void pbvh_update_mask_redraw(PBVH *bvh, PBVHNode **nodes, int totnode, in
       .flag = flag,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, true, totnode);
-  BLI_task_parallel_range(0, totnode, &data, pbvh_update_mask_redraw_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, pbvh_update_mask_redraw_task_cb, &settings);
 }
 
 static void pbvh_update_BB_redraw_task_cb(void *__restrict userdata,
@@ -1189,9 +1188,9 @@ void pbvh_update_BB_redraw(PBVH *bvh, PBVHNode **nodes, int totnode, int flag)
       .flag = flag,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, true, totnode);
-  BLI_task_parallel_range(0, totnode, &data, pbvh_update_BB_redraw_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, pbvh_update_BB_redraw_task_cb, &settings);
 }
 
 static int pbvh_get_buffers_update_flags(PBVH *bvh, bool show_vcol)
@@ -1299,9 +1298,9 @@ static void pbvh_update_draw_buffers(
       .show_vcol = show_vcol,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, true, totnode);
-  BLI_task_parallel_range(0, totnode, &data, pbvh_update_draw_buffer_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, pbvh_update_draw_buffer_cb, &settings);
 }
 
 static int pbvh_flush_bb(PBVH *bvh, PBVHNode *node, int flag)
@@ -2742,13 +2741,10 @@ void pbvh_show_mask_set(PBVH *bvh, bool show_mask)
   bvh->show_mask = show_mask;
 }
 
-void BKE_pbvh_parallel_range_settings(TaskParallelSettings *settings,
+void BKE_pbvh_parallel_range_settings(PBVHParallelSettings *settings,
                                       bool use_threading,
                                       int totnode)
 {
-  const int threaded_limit = 1;
-  BLI_parallel_range_settings_defaults(settings);
-  settings->use_threading = use_threading && (totnode > threaded_limit);
-  settings->min_iter_per_thread = 1;
-  settings->scheduling_mode = TASK_SCHEDULING_DYNAMIC;
+  memset(settings, 0, sizeof(*settings));
+  settings->use_threading = use_threading && totnode > 1;
 }
diff --git a/source/blender/blenkernel/intern/pbvh_parallel.cc b/source/blender/blenkernel/intern/pbvh_parallel.cc
new file mode 100644
index 00000000000..60503c9bad0
--- /dev/null
+++ b/source/blender/blenkernel/intern/pbvh_parallel.cc
@@ -0,0 +1,140 @@
+/*
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
+ */
+
+#include "MEM_guardedalloc.h"
+
+#include "BLI_task.h"
+#include "BLI_threads.h"
+
+#include "BKE_pbvh.h"
+
+#include "atomic_ops.h"
+
+#ifdef WITH_TBB
+
+#  include <tbb/tbb.h>
+
+/* Functor for running TBB parallel_for and parallel_reduce. */
+struct PBVHTask {
+  PBVHParallelRangeFunc func;
+  void *userdata;
+  const PBVHParallelSettings *settings;
+
+  void *userdata_chunk;
+  bool userdata_chunk_free;
+
+  PBVHTask()
+  {
+  }
+
+  PBVHTask(const PBVHTask &other)
+      : func(other.func),
+        userdata(other.userdata),
+        settings(other.settings),
+        userdata_chunk(0),
+        userdata_chunk_free(false)
+  {
+    if (other.userdata_chunk) {
+      userdata_chunk = MEM_mallocN(settings->userdata_chunk_size, "PBVHTask");
+      memcpy(userdata_chunk, other.userdata_chunk, settings->userdata_chunk_size);
+      userdata_chunk_free = true;
+    }
+  }
+
+  PBVHTask(PBVHTask &other, tbb::split) : PBVHTask(other)
+  {
+  }
+
+  ~PBVHTask()
+  {
+    if (userdata_chunk_free) {
+      MEM_freeN(userdata_chunk);
+    }
+  }
+
+  void operator()(const tbb::blocked_range<int> &r) const
+  {
+    TaskParallelTLS tls;
+    tls.thread_id = get_thread_id();
+    tls.userdata_chunk = userdata_chunk;
+    for (int i = r.begin(); i != r.end(); ++i) {
+      func(userdata, i, &tls);
+    }
+  }
+
+  void join(const PBVHTask &other)
+  {
+    settings->func_reduce(userdata, userdata_chunk, other.userdata_chunk);
+  }
+
+  int get_thread_id() const
+  {
+    /* Get a unique thread ID for texture nodes. In the future we should get rid
+     * of the thread ID and change texture evaluation to not require per-thread
+     * storage that can't be efficiently allocated on the stack. */
+    static tbb::enumerable_thread_specific<int> pbvh_thread_id(-1);
+    static int pbvh_thread_id_counter = 0;
+
+    int &thread_id = pbvh_thread_id.local();
+    if (thread_id == -1) {
+      thread_id = atomic_fetch_and_add_int32(&pbvh_thread_id_counter, 1);
+      if (thread_id >= BLENDER_MAX_THREADS) {
+        BLI_assert(!"Maximum number of threads exceeded for sculpting");
+        thread_id = thread_id % BLENDER_MAX_THREADS;
+      }
+    }
+    return thread_id;
+  }
+};
+
+#endif
+
+void BKE_pbvh_parallel_range(const int start,
+                             const int stop,
+                             void *userdata,
+                             PBVHParallelRangeFunc func,
+                             const struct PBVHParallelSettings *settings)
+{
+#ifdef WITH_TBB
+  /* Multithreading. */
+  if (settings->use_threading) {
+    PBVHTask task;
+    task.func = func;
+    task.userdata = userdata;
+    task.settings = settings;
+    task.userdata_chunk = settings->userdata_chunk;
+    task.userdata_chunk_free = false;
+
+    if (settings->func_reduce) {
+      parallel_reduce(tbb::blocked_range<int>(start, stop), task);
+    }
+    else {
+      parallel_for(tbb::blocked_range<int>(start, stop), task);
+    }
+
+    return;
+  }
+#endif
+
+  /* Single threaded. Nothing to reduce as everything is accumulated into the
+   * main userdata chunk directly. */
+  TaskParallelTLS tls;
+  tls.thread_id = 0;
+  tls.userdata_chunk = settings->userdata_chunk;
+  for (int i = start; i < stop; i++) {
+    func(userdata, i, &tls);
+  }
+}
diff --git a/source/blender/editors/sculpt_paint/paint_mask.c b/source/blender/editors/sculpt_paint/paint_mask.c
index a93e55685d2..d160fba4013 100644
--- a/source/blender/editors/sculpt_paint/paint_mask.c
+++ b/source/blender/editors/sculpt_paint/paint_mask.c
@@ -166,9 +166,9 @@ static int mask_flood_fill_exec(bContext *C, wmOperator *op)
       .value = value,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, mask_flood_fill_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, mask_flood_fill_task_cb, &settings);
 
   if (multires) {
     multires_mark_as_modified(depsgraph, ob, MULTIRES_COORDS_MODIFIED);
@@ -343,9 +343,9 @@ bool ED_sculpt_mask_box_select(struct bContext *C, ViewContext *vc, const rcti *
           .clip_planes_final = clip_planes_final,
       };
 
-      TaskParallelSettings settings;
+      PBVHParallelSettings settings;
       BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-      BLI_task_parallel_range(0, totnode, &data, mask_box_select_task_cb, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, mask_box_select_task_cb, &settings);
 
       if (nodes) {
         MEM_freeN(nodes);
@@ -532,9 +532,9 @@ static int paint_mask_gesture_lasso_exec(bContext *C, wmOperator *op)
         data.task_data.mode = mode;
         data.task_data.value = value;
 
-        TaskParallelSettings settings;
+        PBVHParallelSettings settings;
         BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-        BLI_task_parallel_range(0, totnode, &data, mask_gesture_lasso_task_cb, &settings);
+        BKE_pbvh_parallel_range(0, totnode, &data, mask_gesture_lasso_task_cb, &settings);
 
         if (nodes) {
           MEM_freeN(nodes);
diff --git a/source/blender/editors/sculpt_paint/paint_vertex.c b/source/blender/editors/sculpt_paint/paint_vertex.c
index 81e3f04a0d1..77c95c6acb3 100644
--- a/source/blender/editors/sculpt_paint/paint_vertex.c
+++ b/source/blender/editors/sculpt_paint/paint_vertex.c
@@ -2072,9 +2072,9 @@ static void calculate_average_weight(SculptThreadedTaskData *data,
   struct WPaintAverageAccum *accum = MEM_mallocN(sizeof(*accum) * totnode, __func__);
   data->custom_data = accum;
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (data->sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, data, do_wpaint_brush_calc_average_weight_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, data, do_wpaint_brush_calc_average_weight_cb_ex, &settings);
 
   uint accum_len = 0;
   double accum_weight = 0.0;
@@ -2120,22 +2120,22 @@ static void wpaint_paint_leaves(bContext *C,
   data.strength = BKE_brush_weight_get(scene, brush);
 
   /* NOTE: current mirroring code cannot be run in parallel */
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, !(me->editflag & ME_EDIT_MIRROR_X), totnode);
 
   switch ((eBrushWeightPaintTool)brush->weightpaint_tool) {
     case WPAINT_TOOL_AVERAGE:
       calculate_average_weight(&data, nodes, totnode);
-      BLI_task_parallel_range(0, totnode, &data, do_wpaint_brush_draw_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_wpaint_brush_draw_task_cb_ex, &settings);
       break;
     case WPAINT_TOOL_SMEAR:
-      BLI_task_parallel_range(0, totnode, &data, do_wpaint_brush_smear_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_wpaint_brush_smear_task_cb_ex, &settings);
       break;
     case WPAINT_TOOL_BLUR:
-      BLI_task_parallel_range(0, totnode, &data, do_wpaint_brush_blur_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_wpaint_brush_blur_task_cb_ex, &settings);
       break;
     case WPAINT_TOOL_DRAW:
-      BLI_task_parallel_range(0, totnode, &data, do_wpaint_brush_draw_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_wpaint_brush_draw_task_cb_ex, &settings);
       break;
   }
 }
@@ -3126,9 +3126,9 @@ static void calculate_average_color(SculptThreadedTaskData *data,
   struct VPaintAverageAccum *accum = MEM_mallocN(sizeof(*accum) * totnode, __func__);
   data->custom_data = accum;
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, true, totnode);
-  BLI_task_parallel_range(0, totnode, data, do_vpaint_brush_calc_average_color_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, data, do_vpaint_brush_calc_average_color_cb_ex, &settings);
 
   uint accum_len = 0;
   uint accum_value[3] = {0};
@@ -3172,21 +3172,21 @@ static void vpaint_paint_leaves(bContext *C,
       .lcol = (uint *)me->mloopcol,
       .me = me,
   };
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, true, totnode);
   switch ((eBrushVertexPaintTool)brush->vertexpaint_tool) {
     case VPAINT_TOOL_AVERAGE:
       calculate_average_color(&data, nodes, totnode);
-      BLI_task_parallel_range(0, totnode, &data, do_vpaint_brush_draw_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_vpaint_brush_draw_task_cb_ex, &settings);
       break;
     case VPAINT_TOOL_BLUR:
-      BLI_task_parallel_range(0, totnode, &data, do_vpaint_brush_blur_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_vpaint_brush_blur_task_cb_ex, &settings);
       break;
     case VPAINT_TOOL_SMEAR:
-      BLI_task_parallel_range(0, totnode, &data, do_vpaint_brush_smear_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_vpaint_brush_smear_task_cb_ex, &settings);
       break;
     case VPAINT_TOOL_DRAW:
-      BLI_task_parallel_range(0, totnode, &data, do_vpaint_brush_draw_task_cb_ex, &settings);
+      BKE_pbvh_parallel_range(0, totnode, &data, do_vpaint_brush_draw_task_cb_ex, &settings);
       break;
   }
 }
diff --git a/source/blender/editors/sculpt_paint/sculpt.c b/source/blender/editors/sculpt_paint/sculpt.c
index 2d329ff3bca..2bde42c739c 100644
--- a/source/blender/editors/sculpt_paint/sculpt.c
+++ b/source/blender/editors/sculpt_paint/sculpt.c
@@ -423,17 +423,19 @@ static void do_nearest_vertex_get_task_cb(void *__restrict userdata,
   BKE_pbvh_vertex_iter_end;
 }
 
-static void nearest_vertex_get_finalize(void *__restrict userdata, void *__restrict tls)
+static void nearest_vertex_get_reduce(const void *__restrict UNUSED(userdata),
+                                      void *__restrict chunk_join,
+                                      void *__restrict chunk)
 {
-  SculptThreadedTaskData *data = userdata;
-  NearestVertexTLSData *nvtd = tls;
-  if (data->nearest_vertex_index == -1) {
-    data->nearest_vertex_index = nvtd->nearest_vertex_index;
-    data->nearest_vertex_distance_squared = nvtd->nearest_vertex_distance_squared;
+  NearestVertexTLSData *join = chunk_join;
+  NearestVertexTLSData *nvtd = chunk;
+  if (join->nearest_vertex_index == -1) {
+    join->nearest_vertex_index = nvtd->nearest_vertex_index;
+    join->nearest_vertex_distance_squared = nvtd->nearest_vertex_distance_squared;
   }
-  else if (nvtd->nearest_vertex_distance_squared < data->nearest_vertex_distance_squared) {
-    data->nearest_vertex_index = nvtd->nearest_vertex_index;
-    data->nearest_vertex_distance_squared = nvtd->nearest_vertex_distance_squared;
+  else if (nvtd->nearest_vertex_distance_squared < join->nearest_vertex_distance_squared) {
+    join->nearest_vertex_index = nvtd->nearest_vertex_index;
+    join->nearest_vertex_distance_squared = nvtd->nearest_vertex_distance_squared;
   }
 }
 
@@ -460,25 +462,23 @@ static int sculpt_nearest_vertex_get(
       .ob = ob,
       .nodes = nodes,
       .max_distance_squared = max_distance * max_distance,
-      .nearest_vertex_index = -1,
   };
 
   copy_v3_v3(task_data.nearest_vertex_search_co, co);
-  task_data.nearest_vertex_distance_squared = FLT_MAX;
   NearestVertexTLSData nvtd;
   nvtd.nearest_vertex_index = -1;
   nvtd.nearest_vertex_distance_squared = FLT_MAX;
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  settings.func_finalize = nearest_vertex_get_finalize;
+  settings.func_reduce = nearest_vertex_get_reduce;
   settings.userdata_chunk = &nvtd;
   settings.userdata_chunk_size = sizeof(NearestVertexTLSData);
-  BLI_task_parallel_range(0, totnode, &task_data, do_nearest_vertex_get_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &task_data, do_nearest_vertex_get_task_cb, &settings);
 
   MEM_SAFE_FREE(nodes);
 
-  return task_data.nearest_vertex_index;
+  return nvtd.nearest_vertex_index;
 }
 
 static bool is_symmetry_iteration_valid(char i, char symm)
@@ -937,9 +937,9 @@ static void paint_mesh_restore_co(Sculpt *sd, Object *ob)
       .nodes = nodes,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP) && !ss->bm, totnode);
-  BLI_task_parallel_range(0, totnode, &data, paint_mesh_restore_co_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, paint_mesh_restore_co_task_cb, &settings);
 
   MEM_SAFE_FREE(nodes);
 }
@@ -1413,9 +1413,10 @@ static float calc_symmetry_feather(Sculpt *sd, StrokeCache *cache)
  * \{ */
 
 typedef struct AreaNormalCenterTLSData {
-  float private_co[2][3];
-  float private_no[2][3];
-  int private_count[2];
+  /* 0=towards view, 1=flipped */
+  float area_cos[2][3];
+  float area_nos[2][3];
+  int area_count[2];
 } AreaNormalCenterTLSData;
 
 static void calc_area_normal_and_center_task_cb(void *__restrict userdata,
@@ -1425,8 +1426,8 @@ static void calc_area_normal_and_center_task_cb(void *__restrict userdata,
   SculptThreadedTaskData *data = userdata;
   SculptSession *ss = data->ob->sculpt;
   AreaNormalCenterTLSData *anctd = tls->userdata_chunk;
-  float(*area_nos)[3] = data->area_nos;
-  float(*area_cos)[3] = data->area_cos;
+  const bool use_area_nos = data->use_area_nos;
+  const bool use_area_cos = data->use_area_cos;
 
   PBVHVertexIter vd;
   SculptUndoNode *unode = NULL;
@@ -1479,13 +1480,13 @@ static void calc_area_normal_and_center_task_cb(void *__restrict userdata,
         normal_tri_v3(no, UNPACK3(co_tri));
 
         flip_index = (dot_v3v3(ss->cache->view_normal, no) <= 0.0f);
-        if (area_cos) {
-          add_v3_v3(anctd->private_co[flip_index], co);
+        if (use_area_cos) {
+          add_v3_v3(anctd->area_cos[flip_index], co);
         }
-        if (area_nos) {
-          add_v3_v3(anctd->private_no[flip_index], no);
+        if (use_area_nos) {
+          add_v3_v3(anctd->area_nos[flip_index], no);
         }
-        anctd->private_count[flip_index] += 1;
+        anctd->area_count[flip_index] += 1;
       }
     }
   }
@@ -1531,40 +1532,37 @@ static void calc_area_normal_and_center_task_cb(void *__restrict userdata,
 
         flip_index = (dot_v3v3(ss->cache ? ss->cache->view_normal : ss->cursor_view_normal, no) <=
                       0.0f);
-        if (area_cos) {
-          add_v3_v3(anctd->private_co[flip_index], co);
+        if (use_area_cos) {
+          add_v3_v3(anctd->area_cos[flip_index], co);
         }
-        if (area_nos) {
-          add_v3_v3(anctd->private_no[flip_index], no);
+        if (use_area_nos) {
+          add_v3_v3(anctd->area_nos[flip_index], no);
         }
-        anctd->private_count[flip_index] += 1;
+        anctd->area_count[flip_index] += 1;
       }
     }
     BKE_pbvh_vertex_iter_end;
   }
 }
 
-static void calc_area_normal_and_center_finalize(void *__restrict userdata, void *__restrict tls)
+static void calc_area_normal_and_center_reduce(const void *__restrict UNUSED(userdata),
+                                               void *__restrict chunk_join,
+                                               void *__restrict chunk)
 {
-  SculptThreadedTaskData *data = userdata;
-  AreaNormalCenterTLSData *anctd = tls;
-  float(*area_nos)[3] = data->area_nos;
-  float(*area_cos)[3] = data->area_cos;
+  AreaNormalCenterTLSData *join = chunk_join;
+  AreaNormalCenterTLSData *anctd = chunk;
+
   /* for flatten center */
-  if (area_cos) {
-    add_v3_v3(area_cos[0], anctd->private_co[0]);
-    add_v3_v3(area_cos[1], anctd->private_co[1]);
-  }
+  add_v3_v3(join->area_cos[0], anctd->area_cos[0]);
+  add_v3_v3(join->area_cos[1], anctd->area_cos[1]);
 
   /* for area normal */
-  if (area_nos) {
-    add_v3_v3(area_nos[0], anctd->private_no[0]);
-    add_v3_v3(area_nos[1], anctd->private_no[1]);
-  }
+  add_v3_v3(join->area_nos[0], anctd->area_nos[0]);
+  add_v3_v3(join->area_nos[1], anctd->area_nos[1]);
 
   /* weights */
-  data->count[0] += anctd->private_count[0];
-  data->count[1] += anctd->private_count[1];
+  join->area_count[0] += anctd->area_count[0];
+  join->area_count[1] += anctd->area_count[1];
 }
 
 static void calc_area_center(
@@ -1575,11 +1573,6 @@ static void calc_area_center(
   const bool has_bm_orco = ss->bm && sculpt_stroke_is_dynamic_topology(ss, brush);
   int n;
 
-  /* 0=towards view, 1=flipped */
-  float area_cos[2][3] = {{0.0f}};
-
-  int count[2] = {0};
-
   /* Intentionally set 'sd' to NULL since we share logic with vertex paint. */
   SculptThreadedTaskData data = {
       .sd = NULL,
@@ -1588,24 +1581,22 @@ static void calc_area_center(
       .nodes = nodes,
       .totnode = totnode,
       .has_bm_orco = has_bm_orco,
-      .area_cos = area_cos,
-      .area_nos = NULL,
-      .count = count,
+      .use_area_cos = true,
   };
 
   AreaNormalCenterTLSData anctd = {{{0}}};
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  settings.func_finalize = calc_area_normal_and_center_finalize;
+  settings.func_reduce = calc_area_normal_and_center_reduce;
   settings.userdata_chunk = &anctd;
   settings.userdata_chunk_size = sizeof(AreaNormalCenterTLSData);
-  BLI_task_parallel_range(0, totnode, &data, calc_area_normal_and_center_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, calc_area_normal_and_center_task_cb, &settings);
 
   /* for flatten center */
-  for (n = 0; n < ARRAY_SIZE(area_cos); n++) {
-    if (count[n] != 0) {
-      mul_v3_v3fl(r_area_co, area_cos[n], 1.0f / count[n]);
+  for (n = 0; n < ARRAY_SIZE(anctd.area_cos); n++) {
+    if (anctd.area_count[n] != 0) {
+      mul_v3_v3fl(r_area_co, anctd.area_cos[n], 1.0f / anctd.area_count[n]);
       break;
     }
   }
@@ -1633,11 +1624,6 @@ bool sculpt_pbvh_calc_area_normal(const Brush *brush,
   SculptSession *ss = ob->sculpt;
   const bool has_bm_orco = ss->bm && sculpt_stroke_is_dynamic_topology(ss, brush);
 
-  /* 0=towards view, 1=flipped */
-  float area_nos[2][3] = {{0.0f}};
-
-  int count[2] = {0};
-
   /* Intentionally set 'sd' to NULL since this is used for vertex paint too. */
   SculptThreadedTaskData data = {
       .sd = NULL,
@@ -1646,24 +1632,22 @@ bool sculpt_pbvh_calc_area_normal(const Brush *brush,
       .nodes = nodes,
       .totnode = totnode,
       .has_bm_orco = has_bm_orco,
-      .area_cos = NULL,
-      .area_nos = area_nos,
-      .count = count,
+      .use_area_nos = true,
       .any_vertex_sampled = false,
   };
 
   AreaNormalCenterTLSData anctd = {{{0}}};
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, use_threading, totnode);
-  settings.func_finalize = calc_area_normal_and_center_finalize;
+  settings.func_reduce = calc_area_normal_and_center_reduce;
   settings.userdata_chunk = &anctd;
   settings.userdata_chunk_size = sizeof(AreaNormalCenterTLSData);
-  BLI_task_parallel_range(0, totnode, &data, calc_area_normal_and_center_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, calc_area_normal_and_center_task_cb, &settings);
 
   /* for area normal */
-  for (int i = 0; i < ARRAY_SIZE(area_nos); i++) {
-    if (normalize_v3_v3(r_area_no, area_nos[i]) != 0.0f) {
+  for (int i = 0; i < ARRAY_SIZE(anctd.area_nos); i++) {
+    if (normalize_v3_v3(r_area_no, anctd.area_nos[i]) != 0.0f) {
       break;
     }
   }
@@ -1681,12 +1665,6 @@ static void calc_area_normal_and_center(
   const bool has_bm_orco = ss->bm && sculpt_stroke_is_dynamic_topology(ss, brush);
   int n;
 
-  /* 0=towards view, 1=flipped */
-  float area_cos[2][3] = {{0.0f}};
-  float area_nos[2][3] = {{0.0f}};
-
-  int count[2] = {0};
-
   /* Intentionally set 'sd' to NULL since this is used for vertex paint too. */
   SculptThreadedTaskData data = {
       .sd = NULL,
@@ -1695,24 +1673,23 @@ static void calc_area_normal_and_center(
       .nodes = nodes,
       .totnode = totnode,
       .has_bm_orco = has_bm_orco,
-      .area_cos = area_cos,
-      .area_nos = area_nos,
-      .count = count,
+      .use_area_cos = true,
+      .use_area_nos = true,
   };
 
   AreaNormalCenterTLSData anctd = {{{0}}};
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  settings.func_finalize = calc_area_normal_and_center_finalize;
+  settings.func_reduce = calc_area_normal_and_center_reduce;
   settings.userdata_chunk = &anctd;
   settings.userdata_chunk_size = sizeof(AreaNormalCenterTLSData);
-  BLI_task_parallel_range(0, totnode, &data, calc_area_normal_and_center_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, calc_area_normal_and_center_task_cb, &settings);
 
   /* for flatten center */
-  for (n = 0; n < ARRAY_SIZE(area_cos); n++) {
-    if (count[n] != 0) {
-      mul_v3_v3fl(r_area_co, area_cos[n], 1.0f / count[n]);
+  for (n = 0; n < ARRAY_SIZE(anctd.area_cos); n++) {
+    if (anctd.area_count[n] != 0) {
+      mul_v3_v3fl(r_area_co, anctd.area_cos[n], 1.0f / anctd.area_count[n]);
       break;
     }
   }
@@ -1721,8 +1698,8 @@ static void calc_area_normal_and_center(
   }
 
   /* for area normal */
-  for (n = 0; n < ARRAY_SIZE(area_nos); n++) {
-    if (normalize_v3_v3(r_area_no, area_nos[n]) != 0.0f) {
+  for (n = 0; n < ARRAY_SIZE(anctd.area_nos); n++) {
+    if (normalize_v3_v3(r_area_no, anctd.area_nos[n]) != 0.0f) {
       break;
     }
   }
@@ -2824,7 +2801,7 @@ static void smooth(Sculpt *sd,
         .strength = strength,
     };
 
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
 
     switch (type) {
@@ -2842,16 +2819,16 @@ static void smooth(Sculpt *sd,
 
         settings.userdata_chunk = data_chunk;
         settings.userdata_chunk_size = size;
-        BLI_task_parallel_range(0, totnode, &data, do_smooth_brush_multires_task_cb_ex, &settings);
+        BKE_pbvh_parallel_range(0, totnode, &data, do_smooth_brush_multires_task_cb_ex, &settings);
 
         MEM_freeN(data_chunk);
         break;
       }
       case PBVH_FACES:
-        BLI_task_parallel_range(0, totnode, &data, do_smooth_brush_mesh_task_cb_ex, &settings);
+        BKE_pbvh_parallel_range(0, totnode, &data, do_smooth_brush_mesh_task_cb_ex, &settings);
         break;
       case PBVH_BMESH:
-        BLI_task_parallel_range(0, totnode, &data, do_smooth_brush_bmesh_task_cb_ex, &settings);
+        BKE_pbvh_parallel_range(0, totnode, &data, do_smooth_brush_bmesh_task_cb_ex, &settings);
         break;
     }
 
@@ -2883,10 +2860,10 @@ static void bmesh_topology_rake(
         .nodes = nodes,
         .strength = factor,
     };
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
 
-    BLI_task_parallel_range(0, totnode, &data, do_topology_rake_bmesh_task_cb_ex, &settings);
+    BKE_pbvh_parallel_range(0, totnode, &data, do_topology_rake_bmesh_task_cb_ex, &settings);
   }
 }
 
@@ -2940,9 +2917,9 @@ static void do_mask_brush_draw(Sculpt *sd, Object *ob, PBVHNode **nodes, int tot
       .nodes = nodes,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_mask_brush_draw_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_mask_brush_draw_task_cb_ex, &settings);
 }
 
 static void do_mask_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode)
@@ -3027,9 +3004,9 @@ static void do_draw_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode)
       .offset = offset,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_draw_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_draw_brush_task_cb_ex, &settings);
 }
 
 static void do_draw_sharp_brush_task_cb_ex(void *__restrict userdata,
@@ -3103,9 +3080,9 @@ static void do_draw_sharp_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int to
       .offset = offset,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_draw_sharp_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_draw_sharp_brush_task_cb_ex, &settings);
 }
 
 /**
@@ -3219,9 +3196,9 @@ static void do_crease_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnod
       .flippedbstrength = flippedbstrength,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_crease_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_crease_brush_task_cb_ex, &settings);
 }
 
 static void do_pinch_brush_task_cb_ex(void *__restrict userdata,
@@ -3281,9 +3258,9 @@ static void do_pinch_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode
       .nodes = nodes,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_pinch_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_pinch_brush_task_cb_ex, &settings);
 }
 
 static void do_grab_brush_task_cb_ex(void *__restrict userdata,
@@ -3353,9 +3330,9 @@ static void do_grab_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode)
       .grab_delta = grab_delta,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_grab_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_grab_brush_task_cb_ex, &settings);
 }
 
 /* Regularized Kelvinlets: Sculpting Brushes based on Fundamental Solutions of Elasticity
@@ -3606,9 +3583,9 @@ static void do_elastic_deform_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, in
       .grab_delta = grab_delta,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_elastic_deform_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_elastic_deform_brush_task_cb_ex, &settings);
 }
 
 static void do_pose_brush_task_cb_ex(void *__restrict userdata,
@@ -3695,14 +3672,14 @@ static void do_pose_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode)
       .transform_trans_inv = transform_trans_inv,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_pose_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_pose_brush_task_cb_ex, &settings);
 }
 
 typedef struct PoseGrowFactorTLSData {
   float pos_avg[3];
-  int tot_pos_avg;
+  int pos_count;
 } PoseGrowFactorTLSData;
 
 static void pose_brush_grow_factor_task_cb_ex(void *__restrict userdata,
@@ -3731,7 +3708,7 @@ static void pose_brush_grow_factor_task_cb_ex(void *__restrict userdata,
       data->pose_factor[vd.index] = max;
       if (check_vertex_pivot_symmetry(vd.co, active_co, symm)) {
         add_v3_v3(gftd->pos_avg, vd.co);
-        gftd->tot_pos_avg++;
+        gftd->pos_count++;
       }
     }
   }
@@ -3739,12 +3716,14 @@ static void pose_brush_grow_factor_task_cb_ex(void *__restrict userdata,
   BKE_pbvh_vertex_iter_end;
 }
 
-static void pose_brush_grow_factor_finalize(void *__restrict userdata, void *__restrict tls)
+static void pose_brush_grow_factor_reduce(const void *__restrict UNUSED(userdata),
+                                          void *__restrict chunk_join,
+                                          void *__restrict chunk)
 {
-  SculptThreadedTaskData *data = userdata;
-  PoseGrowFactorTLSData *gftd = tls;
-  add_v3_v3(data->tot_pos_avg, gftd->pos_avg);
-  data->tot_pos_count += gftd->tot_pos_avg;
+  PoseGrowFactorTLSData *join = chunk_join;
+  PoseGrowFactorTLSData *gftd = chunk;
+  add_v3_v3(join->pos_avg, gftd->pos_avg);
+  join->pos_count += gftd->pos_count;
 }
 
 /* Grow the factor until its boundary is near to the offset pose origin */
@@ -3763,12 +3742,12 @@ static void sculpt_pose_grow_pose_factor(
       .totnode = totnode,
       .pose_factor = pose_factor,
   };
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   PoseGrowFactorTLSData gftd;
-  gftd.tot_pos_avg = 0;
+  gftd.pos_count = 0;
   zero_v3(gftd.pos_avg);
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  settings.func_finalize = pose_brush_grow_factor_finalize;
+  settings.func_reduce = pose_brush_grow_factor_reduce;
   settings.userdata_chunk = &gftd;
   settings.userdata_chunk_size = sizeof(PoseGrowFactorTLSData);
 
@@ -3776,15 +3755,13 @@ static void sculpt_pose_grow_pose_factor(
   float prev_len = FLT_MAX;
   data.prev_mask = MEM_mallocN(sculpt_vertex_count_get(ss) * sizeof(float), "prev mask");
   while (grow_next_iteration) {
-    zero_v3(data.tot_pos_avg);
-    data.tot_pos_count = 0;
     zero_v3(gftd.pos_avg);
-    gftd.tot_pos_avg = 0;
+    gftd.pos_count = 0;
     memcpy(data.prev_mask, pose_factor, sculpt_vertex_count_get(ss) * sizeof(float));
-    BLI_task_parallel_range(0, totnode, &data, pose_brush_grow_factor_task_cb_ex, &settings);
-    if (data.tot_pos_count != 0) {
-      mul_v3_fl(data.tot_pos_avg, 1.0f / (float)data.tot_pos_count);
-      float len = len_v3v3(data.tot_pos_avg, pose_origin);
+    BKE_pbvh_parallel_range(0, totnode, &data, pose_brush_grow_factor_task_cb_ex, &settings);
+    if (gftd.pos_count != 0) {
+      mul_v3_fl(gftd.pos_avg, 1.0f / (float)gftd.pos_count);
+      float len = len_v3v3(gftd.pos_avg, pose_origin);
       if (len < prev_len) {
         prev_len = len;
         grow_next_iteration = true;
@@ -3953,9 +3930,9 @@ static void sculpt_pose_brush_init(
 
   /* Smooth the pose brush factor for cleaner deformation */
   for (int i = 0; i < 4; i++) {
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-    BLI_task_parallel_range(0, totnode, &data, pose_brush_init_task_cb_ex, &settings);
+    BKE_pbvh_parallel_range(0, totnode, &data, pose_brush_init_task_cb_ex, &settings);
   }
 
   MEM_SAFE_FREE(nodes);
@@ -4023,9 +4000,9 @@ static void do_nudge_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode
       .cono = cono,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_nudge_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_nudge_brush_task_cb_ex, &settings);
 }
 
 static void do_snake_hook_brush_task_cb_ex(void *__restrict userdata,
@@ -4144,9 +4121,9 @@ static void do_snake_hook_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int to
       .grab_delta = grab_delta,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_snake_hook_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_snake_hook_brush_task_cb_ex, &settings);
 }
 
 static void do_thumb_brush_task_cb_ex(void *__restrict userdata,
@@ -4216,9 +4193,9 @@ static void do_thumb_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode
       .cono = cono,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_thumb_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_thumb_brush_task_cb_ex, &settings);
 }
 
 static void do_rotate_brush_task_cb_ex(void *__restrict userdata,
@@ -4289,9 +4266,9 @@ static void do_rotate_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnod
       .angle = angle,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_rotate_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_rotate_brush_task_cb_ex, &settings);
 }
 
 static void do_layer_brush_task_cb_ex(void *__restrict userdata,
@@ -4386,9 +4363,9 @@ static void do_layer_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode
   };
   BLI_mutex_init(&data.mutex);
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_layer_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_layer_brush_task_cb_ex, &settings);
 
   BLI_mutex_end(&data.mutex);
 }
@@ -4454,9 +4431,9 @@ static void do_inflate_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totno
       .nodes = nodes,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_inflate_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_inflate_brush_task_cb_ex, &settings);
 }
 
 static void calc_sculpt_plane(
@@ -4663,9 +4640,9 @@ static void do_flatten_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totno
       .area_co = area_co,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_flatten_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_flatten_brush_task_cb_ex, &settings);
 }
 
 static void do_clay_brush_task_cb_ex(void *__restrict userdata,
@@ -4761,9 +4738,9 @@ static void do_clay_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode)
       .area_co = area_co,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_clay_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_clay_brush_task_cb_ex, &settings);
 }
 
 static void do_clay_strips_brush_task_cb_ex(void *__restrict userdata,
@@ -4891,9 +4868,9 @@ static void do_clay_strips_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int t
       .mat = mat,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_clay_strips_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_clay_strips_brush_task_cb_ex, &settings);
 }
 
 static void do_fill_brush_task_cb_ex(void *__restrict userdata,
@@ -4984,9 +4961,9 @@ static void do_fill_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode)
       .area_co = area_co,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_fill_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_fill_brush_task_cb_ex, &settings);
 }
 
 static void do_scrape_brush_task_cb_ex(void *__restrict userdata,
@@ -5076,9 +5053,9 @@ static void do_scrape_brush(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnod
       .area_co = area_co,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_scrape_brush_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_scrape_brush_task_cb_ex, &settings);
 }
 
 static void do_gravity_task_cb_ex(void *__restrict userdata,
@@ -5145,9 +5122,9 @@ static void do_gravity(Sculpt *sd, Object *ob, PBVHNode **nodes, int totnode, fl
       .offset = offset,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-  BLI_task_parallel_range(0, totnode, &data, do_gravity_task_cb_ex, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, do_gravity_task_cb_ex, &settings);
 }
 
 void sculpt_vertcos_to_key(Object *ob, KeyBlock *kb, const float (*vertCos)[3])
@@ -5319,9 +5296,9 @@ static void do_brush_action(Sculpt *sd, Object *ob, Brush *brush, UnifiedPaintSe
         .nodes = nodes,
     };
 
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-    BLI_task_parallel_range(0, totnode, &task_data, do_brush_action_task_cb, &settings);
+    BKE_pbvh_parallel_range(0, totnode, &task_data, do_brush_action_task_cb, &settings);
 
     if (sculpt_brush_needs_normal(ss, brush)) {
       update_sculpt_normal(sd, ob, nodes, totnode);
@@ -5539,9 +5516,9 @@ static void sculpt_combine_proxies(Sculpt *sd, Object *ob)
         .nodes = nodes,
     };
 
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-    BLI_task_parallel_range(0, totnode, &data, sculpt_combine_proxies_task_cb, &settings);
+    BKE_pbvh_parallel_range(0, totnode, &data, sculpt_combine_proxies_task_cb, &settings);
   }
 
   MEM_SAFE_FREE(nodes);
@@ -5629,9 +5606,9 @@ static void sculpt_flush_stroke_deform(Sculpt *sd, Object *ob, bool is_proxy_use
         .vertCos = vertCos,
     };
 
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-    BLI_task_parallel_range(0, totnode, &data, sculpt_flush_stroke_deform_task_cb, &settings);
+    BKE_pbvh_parallel_range(0, totnode, &data, sculpt_flush_stroke_deform_task_cb, &settings);
 
     if (vertCos) {
       sculpt_vertcos_to_key(ob, ss->shapekey_active, vertCos);
@@ -8216,10 +8193,10 @@ static void sculpt_filter_cache_init(Object *ob, Sculpt *sd)
       .nodes = ss->filter_cache->nodes,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(
       &settings, (sd->flags & SCULPT_USE_OPENMP), ss->filter_cache->totnode);
-  BLI_task_parallel_range(
+  BKE_pbvh_parallel_range(
       0, ss->filter_cache->totnode, &data, filter_cache_init_task_cb, &settings);
 }
 
@@ -8415,10 +8392,10 @@ static int sculpt_mesh_filter_modal(bContext *C, wmOperator *op, const wmEvent *
       .filter_strength = filter_strength,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(
       &settings, (sd->flags & SCULPT_USE_OPENMP), ss->filter_cache->totnode);
-  BLI_task_parallel_range(0, ss->filter_cache->totnode, &data, mesh_filter_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, ss->filter_cache->totnode, &data, mesh_filter_task_cb, &settings);
 
   if (ss->deform_modifiers_active || ss->shapekey_active) {
     sculpt_flush_stroke_deform(sd, ob, true);
@@ -8692,9 +8669,9 @@ static int sculpt_mask_filter_exec(bContext *C, wmOperator *op)
         .prev_mask = prev_mask,
     };
 
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-    BLI_task_parallel_range(0, totnode, &data, mask_filter_task_cb, &settings);
+    BKE_pbvh_parallel_range(0, totnode, &data, mask_filter_task_cb, &settings);
 
     if (ELEM(filter_type, MASK_FILTER_GROW, MASK_FILTER_SHRINK)) {
       MEM_freeN(prev_mask);
@@ -8803,13 +8780,14 @@ static void dirty_mask_compute_range_task_cb(void *__restrict userdata,
   BKE_pbvh_vertex_iter_end;
 }
 
-static void dirty_mask_compute_range_finalize(void *__restrict userdata, void *__restrict tls)
+static void dirty_mask_compute_range_reduce(const void *__restrict UNUSED(userdata),
+                                            void *__restrict chunk_join,
+                                            void *__restrict chunk)
 {
-  SculptThreadedTaskData *data = userdata;
-  DirtyMaskRangeData *range = tls;
-
-  data->dirty_mask_min = min_ff(range->min, data->dirty_mask_min);
-  data->dirty_mask_max = max_ff(range->max, data->dirty_mask_max);
+  DirtyMaskRangeData *join = chunk_join;
+  DirtyMaskRangeData *range = chunk;
+  join->min = min_ff(range->min, join->min);
+  join->max = max_ff(range->max, join->max);
 }
 
 static void dirty_mask_apply_task_cb(void *__restrict userdata,
@@ -8880,8 +8858,6 @@ static int sculpt_dirty_mask_exec(bContext *C, wmOperator *op)
       .sd = sd,
       .ob = ob,
       .nodes = nodes,
-      .dirty_mask_min = FLT_MAX,
-      .dirty_mask_max = -FLT_MAX,
       .dirty_mask_dirty_only = RNA_boolean_get(op->ptr, "dirty_only"),
   };
   DirtyMaskRangeData range = {
@@ -8889,15 +8865,17 @@ static int sculpt_dirty_mask_exec(bContext *C, wmOperator *op)
       .max = -FLT_MAX,
   };
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
 
-  settings.func_finalize = dirty_mask_compute_range_finalize;
+  settings.func_reduce = dirty_mask_compute_range_reduce;
   settings.userdata_chunk = &range;
   settings.userdata_chunk_size = sizeof(DirtyMaskRangeData);
 
-  BLI_task_parallel_range(0, totnode, &data, dirty_mask_compute_range_task_cb, &settings);
-  BLI_task_parallel_range(0, totnode, &data, dirty_mask_apply_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, totnode, &data, dirty_mask_compute_range_task_cb, &settings);
+  data.dirty_mask_min = range.min;
+  data.dirty_mask_max = range.max;
+  BKE_pbvh_parallel_range(0, totnode, &data, dirty_mask_apply_task_cb, &settings);
 
   MEM_SAFE_FREE(nodes);
 
@@ -9056,10 +9034,10 @@ static int sculpt_mask_expand_modal(bContext *C, wmOperator *op, const wmEvent *
     int smooth_iterations = RNA_int_get(op->ptr, "smooth_iterations");
     BKE_sculpt_update_object_for_edit(depsgraph, ob, true, false);
     for (int i = 0; i < smooth_iterations; i++) {
-      TaskParallelSettings settings;
+      PBVHParallelSettings settings;
       BKE_pbvh_parallel_range_settings(
           &settings, (sd->flags & SCULPT_USE_OPENMP), ss->filter_cache->totnode);
-      BLI_task_parallel_range(0, ss->filter_cache->totnode, &data, mask_filter_task_cb, &settings);
+      BKE_pbvh_parallel_range(0, ss->filter_cache->totnode, &data, mask_filter_task_cb, &settings);
     }
 
     /* Pivot position */
@@ -9125,10 +9103,10 @@ static int sculpt_mask_expand_modal(bContext *C, wmOperator *op, const wmEvent *
         .mask_expand_invert_mask = RNA_boolean_get(op->ptr, "invert"),
         .mask_expand_keep_prev_mask = RNA_boolean_get(op->ptr, "keep_previous_mask"),
     };
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(
         &settings, (sd->flags & SCULPT_USE_OPENMP), ss->filter_cache->totnode);
-    BLI_task_parallel_range(0, ss->filter_cache->totnode, &data, sculpt_expand_task_cb, &settings);
+    BKE_pbvh_parallel_range(0, ss->filter_cache->totnode, &data, sculpt_expand_task_cb, &settings);
     ss->filter_cache->mask_update_current_it = mask_expand_update_it;
   }
 
@@ -9277,10 +9255,10 @@ static int sculpt_mask_expand_invoke(bContext *C, wmOperator *op, const wmEvent
       .mask_expand_invert_mask = RNA_boolean_get(op->ptr, "invert"),
       .mask_expand_keep_prev_mask = RNA_boolean_get(op->ptr, "keep_previous_mask"),
   };
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(
       &settings, (sd->flags & SCULPT_USE_OPENMP), ss->filter_cache->totnode);
-  BLI_task_parallel_range(0, ss->filter_cache->totnode, &data, sculpt_expand_task_cb, &settings);
+  BKE_pbvh_parallel_range(0, ss->filter_cache->totnode, &data, sculpt_expand_task_cb, &settings);
 
   const char *status_str = TIP_(
       "Move the mouse to expand the mask from the active vertex. LBM: confirm mask, ESC/RMB: "
@@ -9594,10 +9572,10 @@ void ED_sculpt_update_modal_transform(struct bContext *C)
     mul_m4_m4m4(data.transform_mats[i], pivot_mat, data.transform_mats[i]);
   }
 
-  TaskParallelSettings settings;
+  PBVHParallelSettings settings;
   BKE_pbvh_parallel_range_settings(
       &settings, (sd->flags & SCULPT_USE_OPENMP), ss->filter_cache->totnode);
-  BLI_task_parallel_range(
+  BKE_pbvh_parallel_range(
       0, ss->filter_cache->totnode, &data, sculpt_transform_task_cb, &settings);
 
   if (ss->deform_modifiers_active || ss->shapekey_active) {
diff --git a/source/blender/editors/sculpt_paint/sculpt_intern.h b/source/blender/editors/sculpt_paint/sculpt_intern.h
index 7f6b097060c..93e4a777569 100644
--- a/source/blender/editors/sculpt_paint/sculpt_intern.h
+++ b/source/blender/editors/sculpt_paint/sculpt_intern.h
@@ -195,10 +195,8 @@ typedef struct SculptThreadedTaskData {
   int filter_type;
   float filter_strength;
 
-  /* 0=towards view, 1=flipped */
-  float (*area_cos)[3];
-  float (*area_nos)[3];
-  int *count;
+  bool use_area_cos;
+  bool use_area_nos;
   bool any_vertex_sampled;
 
   float *prev_mask;
@@ -208,13 +206,8 @@ typedef struct SculptThreadedTaskData {
   float *pose_factor;
   float (*transform_rot)[4], (*transform_trans)[4], (*transform_trans_inv)[4];
 
-  float tot_pos_avg[3];
-  int tot_pos_count;
-
   float max_distance_squared;
   float nearest_vertex_search_co[3];
-  int nearest_vertex_index;
-  float nearest_vertex_distance_squared;
 
   int mask_expand_update_it;
   bool mask_expand_invert_mask;
diff --git a/source/blender/editors/sculpt_paint/sculpt_undo.c b/source/blender/editors/sculpt_paint/sculpt_undo.c
index ef064907d94..5d95cc80280 100644
--- a/source/blender/editors/sculpt_paint/sculpt_undo.c
+++ b/source/blender/editors/sculpt_paint/sculpt_undo.c
@@ -360,9 +360,9 @@ static void sculpt_undo_bmesh_restore_generic(bContext *C,
 
     BKE_pbvh_search_gather(ss->pbvh, NULL, NULL, &nodes, &totnode);
 
-    TaskParallelSettings settings;
+    PBVHParallelSettings settings;
     BKE_pbvh_parallel_range_settings(&settings, (sd->flags & SCULPT_USE_OPENMP), totnode);
-    BLI_task_parallel_range(
+    BKE_pbvh_parallel_range(
         0, totnode, nodes, sculpt_undo_bmesh_restore_generic_task_cb, &settings);
 
     if (nodes) {